LCG Production Services - LCG Grid Deployment --- Work Log 2005 here --- Work Log 2006 here

Production Services Work Log (2007)

  • 2007-12-17: Patches #1491 and #1521 installed on wms103, wms105, wms109, wms110, wms112 and wms113.

  • 2007-12-14: wms107 back in production (fans replaced).

  • 2007-12-14: Patches #1491 and #1521 installed on wms106, wms108 and wms111.

  • 2007-12-13: CMS VO removed from the configuration of wms110 (this node is an Atlas node).

  • 2007-12-13: Actuator used to restart the workload manager disabled on the CMS WMS nodes (wms102, wms104, wms107 and wms115).

  • 2007-12-13: Installation from scratch of a new LCG RB rb130 for CMS (in replacement of rb128 which is still down).

  • 2007-12-13: Installation from scratch of wms115 with patches #1491 and #1521.

  • 2007-12-12: Installation of the google pertools packages on wms102 (google-perftools-0.8-1.i386.rpm and google-perftools-devel-0.8-1.i386.rpm). Line /usr/lib/libtcmalloc.so added in the file /opt/glite/etc/init.d/glite-wms-wm.

  • 2007-12-12: Reinstallation from scratch of wms101 with patches #1491 and #1521.

  • 2007-12-11: Reinstallation from scratch of wms107 with patches #1491 and #1521.

  • 2007-12-07: Reinstallation from scratch of wms104 with patches #1491 and #1521.

  • 2007-12-03: Patches #1491 and #1521 installed on wms102. CMS is actually doing some tests with this new version of the middleware.

  • 2007-12-03: Kernel upgrade on monb002.

  • 2007-11-29: VO geant4 supported on rb129.

  • 2007-11-28: LCG RB nodes fully quattorized (package apt removed and no more used for middleware installation).

  • 2007-11-28: gLite WMS nodes wms101 to wms114 fully quattorized (package apt removed and no more used for middleware installation).

  • 2007-11-28: rpmverify installed on LB nodes lb101 to lb103.

  • 2007-11-28: gLite LB nodes lb101 to lb103 fully quattorized (package apt removed and no more used for middleware installation).

  • 2007-11-27: Installation and configuration of the new DPM node lxdpm104 for SAM.

  • 2007-11-22: LCG RB rb128 for CMS down (hardware problem).

  • 2007-11-22: CAs upgrade on LCG RBs, gLite WMS and LB nodes.

  • 2007-11-20: Kernel upgrade on monb003.

  • 2007-11-19: RAID_TW_CTLR alarm on wms101.

  • 2007-11-19: nmi_received alarm on bdii109 (BDII for SAM).

  • 2007-11-15: Update 36 for gLite 3.0 done on all the gLite WMS and LB nodes.

  • 2007-11-13: Update 36 for gLite 3.0 done on all the LCG RB nodes.

  • 2007-11-12: Hardware problem on wms101 (RAID controller). Need to make a backup of the /data02 partition. All the services have been stopped. Back in production in the evening.

  • 2007-11-11: Hardware problem on rb106. It has been fixed by rebooting the node.

  • 2007-11-10: Network server in infinite loop on the LCG RB dedicated to Alice rb105 and rb114. The service has been restarted.

  • 2007-11-10: Sandbox partition is growing quickly due to some large sandboxes produced by a user. The user has been contacted directly.

  • 2007-11-06: MySQL port opened on lb102 and lb103 for the Real Time Monitor (RTM) tool.

  • 2007-11-06: wms114 for SAM put in production.

  • 2007-11-06: Middleware upgrade on all the LCG RB nodes.

  • 2007-11-05: wms103 in drain mode. Back in production in the evening.

  • 2007-11-01: wms102 back in production.

  • 2007-10-31: wms102 in drain mode in the evening.

  • 2007-10-30: rb108 removed from production. This node will be renamed as wms114.

  • 2007-10-30: High load on rb108 due to huge number of gridftp connections. Grid user contacted.

  • 2007-10-29: wms103 back in production.

  • 2007-10-27: wms103 put in drain mode.

  • 2007-10-27: wms110 and wms101 back in production.

  • 2007-10-26: New SAM WMS wms113 put in production.

  • 2007-10-26: wms107 put in drain mode. Back in production during the evening.

  • 2007-10-25: wms101 and wms110 put in drain mode.

  • 2007-10-25: wms104 back in production.

  • 2007-10-24: UI lxn1179 upgraded to version 3.1.0-2.

  • 2007-10-24: wms112 is assigned to SAM and replace rb118 which has been removed from production. This node supports ops, dteam and few user DNs. Thresholds used by the wmproxy have been modified to support higher loads.

  • 2007-10-24: wms107 back in production.

  • 2007-10-24: Partition tmp full on lb102. Developers contacted. These are dump files from LB server, containing data of purged jobs. I modified manually the file /opt/glite/etc/profile.d/grid-env.sh on lb102 and lb103 by adding the variable GLITE_LB_EXPORT_DUMPDIR_KEEP set to "" (it can be a path, eg. /var/glite/LB-dump).

  • 2007-10-23: rb118 (SAM WMS 3.0) removed from production and replaced by wms112.

  • 2007-10-22: wms104 put in drain mode and AFS configuration files updated accordingly. The input.fl file should be emptied in two days.

  • 2007-10-21: wms107 put in drain mode and AFS configuration files updated accordingly. The input.fl file should be emptied in two days.

  • 2007-10-16: As requested by CMS for CSA07, the following modifications have been done on some WMS 3.1 actually in production, with the agreement of Alice, LHCb and Atlas:
    • CMS enabled on wms101 (owned by LHCb).
    • CMS enabled on wms103 (owned by Alice).
    • wms110 given to CMS exclusively (owned by Atlas).

  • 2007-10-16: clean-up done by Rolandas on wms102 and wms104 (condor queue and input.fl file cleaned). These two machines are back in production now.

  • 2007-10-15: wms104 in drain mode for one week.

  • 2007-10-12: CA upgrade to version 1-17 on all gLite WMS and LB nodes.

  • 2007-10-12: Reduce the maximum Sandbox output file to 50MB on rb128 as requested by CMS (add line MaxOutputSandboxSize = 50000000; in file /opt/edg/etc/edg_wl.conf) and file /opt/lcg/etc/cleanup-sandboxes.conf modified to delete all files in sandbox directory older than 7 days or with size greater than 50MB (variables IDLE_MAX and SIZE_MAX).

  • 2007-10-12: Sandbox partition almost full on rb128. Some clean-up has been made to free some space.

  • 2007-10-11: wms107 down during 1 hour. Probably a problem with the RAID controller. Will be investigated by the sysadmins.

  • 2007-10-11: CA upgrade to version 1-17 on all LCG RB nodes.

  • 2007-10-11: Raid controller changed on rb128 this morning.

  • 2007-10-09: spma triggered on all the WMS/LB/RB nodes. A new kernel (2.4.21-51) has been installed. A reboot of all the nodes needs to be planned.

  • 2007-10-09: Problem with the script glite-lb-export.sh (package glite-lb-client) on lb102. This fails with "Error running the edg_wll_Purge() Transport endpoint is not connected ((null))". Under investigation by the developers.

  • 2007-10-08: Package glite-lb-client manually added on lb102 and lb103.

  • 2007-10-08: All the WMS nodes are now using lb103 as LB node.

  • 2007-10-08: Package lcg-fw upgraded to version 1-8 on the lxdpm10x nodes.

  • 2007-10-08: Hardware problem temporarily fixed on rb128 but a new intervention needs to be planned next week to change the raid controller once again. Note that after the reboot, the lemon-host-check command did not work (No monitoring agent process running). This is under investigation.

  • 2007-10-04: New gLite WMS 3.1 nodes put in production:
    • wms109: Alice
    • wms110: Atlas
    • wms111: Atlas
    • wms112: for test only (support dteam and ops).

  • 2007-10-04: New LB node lb103 put in production. This LB will be used by wms109 to lb112.

  • 2007-10-03: Hardware problem on rb128 (RAID controller). Intervention planned tomorrow (Thursday 04 October). Services stopped until the end of the intervention and broadcast sent.

  • 2007-09-21: AFS UI 3.1 lxn1179 put in production (SLC4 OS).

  • 2007-09-19: wms108 put in production. This node supports the VOs geant4, gear, unosat, eela and sixt.

  • 2007-09-19: Add some pool accounts for VO dteam on lxdpm102.

  • 2007-09-19: Classic SE lxn1183 removed from production.

  • 2007-09-18: New LCG RB rb129 for Alice put in production.

  • 2007-09-14: wms102 in drain mode because of large backlog on the workload manager side (file input.fl). See file .drain in _/var/glite.

  • 2007-09-13: Alarm var_full on lxdpm101. The problem is that the MySQL database (actual size = 2.5GB) is using partition /var of size 5GB. The MySQL database must be migrated on another partition (/data01 for example).

  • 2007-09-12: wms107 put in production for CMS. wms108 used for tests only for the time being.

  • 2007-09-10: rb102 and rb109 (CMS WMS) removed from production. rb102 will be renamed as wms107 and reinstalled with the new gLite middleware, and rb109 will be renamed as wms108 and used for tests.

  • 2007-08-31: Middleware upgrade on all LCG RBs in production.

  • 2007-08-31: Middleware upgrade on lxn1194 (GD UI), lxn1183 (+ kernel upgrade).

  • 2007-08-31: Load limit values (thresholds) restored to initial value 10 in /opt/glite/etc/glite_wms.conf on wms102 and wms104.

  • 2007-08-28: Fix bug #29110 on the wms1xx nodes (the workaround is to define the variable MYPROXY_TCP_PORT_RANGE in file /etc/glite/profile.d/glite_setenv.sh).

  • 2007-08-22: Reinstallation from scratch of wms101 (LHCb), wms103 (Alice), wms104 (CMS) and wms105 (Atlas) with the 3.0.2 WMS Checkpoint release (patch 1251).

  • 2007-08-21: Reinstallation from scratch of wms106 (LHCb) with the 3.0.2 WMS Checkpoint release (patch 1251).

  • 2007-08-15: Reinstallation from scratch of lb102 (gLite LB shared by all the VOs) and wms102 (gLite WMS for CMS) with the 3.0.2 WMS Checkpoint release (patch 1251).

  • 2007-08-06: dg20logd.* files with DG.JOBID="https://rb124.cern.ch:7772/..." moved to /var/tmp/bad_1/ on rb124.

  • 2007-08-06: smart_selftest alarm on rb101, rb102, rb103, rb107, rb109.

  • 2007-08-06: disabled xinetd_wrong alarm on rb201 - xinetd package is not installed.

  • 2007-08-02: smart_selftest alarm on rb102 and rb103.

  • 2007-08-01: rb118 back in production after BBU replacement.

  • 2007-07-31: rb118 removed from production because of a problem with the BBU (Battery Backup Unit). Machine switched off in the afternoon.

  • 2007-07-27: wms106 reinstalled from scratch. No middleware installed for the time being. Machine dedicated to LHCb.

  • 2007-07-27: LCG RB rb105.cern.ch dedicated to Alice back in production.

  • 2007-07-27: Kernel upgrade for all the SAM nodes (RBs, WMS, UIs, MONBOX, DPM).

  • 2007-07-27: rb112 (Testing gLite WMS 3.1 for LHCb) removed from production. This machine will be renamed as wms106 and put in cluster gridwms.

  • 2007-07-25: rb105 removed from production because of some problems with two of the RAID disks.

  • 2007-07-25: locallogger restarted on wms104 (backlog files in /var/glite/log/ directory). Bug already know by the developers.

  • 2007-07-25: Kernel upgrade for wms101 (LHCb WMS), rb114 and rb123 (LHCb LCG RBs).

  • 2007-07-25: Kernel upgrade for rb105, rb120 (Alice LCG RBs).

  • 2007-07-24: Kernel upgrade for wms103, rb116 (Alice gLite WMS).

  • 2007-07-23: Kernel upgrade for monb004 and monb005

  • 2007-07-19: New gLite wms 3.1 put in production: wms103, wms104 and wms105.

  • 2007-07-18: Alarm raid_tw triggered on wms101. Machine put in maintenance by the sysadmins (vendor call) but the services are still running. The intervention should be transparent.

  • 2007-07-16: rb126 (Atlas) removed from production this afternoon. This node will be reinstalled from scratch with the new 3.1 middleware and will be renamed as wms105.

  • 2007-07-12: rb125 (CMS) removed from production this afternoon. This node will be reinstalled from scratch with the new 3.1 middleware and will be renamed as wms104.

  • 2007-07-12: rb111 (Alice) removed from production this afternoon. This node will be reinstalled from scratch with the new 3.1 middleware and will be renamed as wms103.

  • 2007-07-10: CA update to version 1.15-1 on RB nodes and WMS nodes.

  • 2007-07-10: job submission to wms101 hangs. Restarting the service related to the wmproxy solved the problem.

  • 2007-07-10: wmproxy service restarted on wms101 because of lot of httpd processes running on it, causing the job submission to hang (see details in Di's email).

  • 2007-07-10: All the gdrbxx nodes have been removed from cluster GD-RB.

  • 2007-07-10: no_contact alarm on rb107. Machine rebooted by the sysadmins. Back in production.

  • 2007-07-07: Segmentation fault for the workload wanager service running on wms101. Developers contacted. Seems to be related to bug #26857: the "max-rank" selection algorithm for collection under some particular circumstances does not work properly).

  • 2007-07-02: Installation and configuration of Lemon sensors on wms101 and lb101. Operational procedures updated for the two new cluster gridwms and gridlb.

  • 2007-07-02: Middleware upgrade (update 27 for gLite 3.0). Nodes types involved: lcg-RBs.

  • 2007-07-01: rb123 down this afternoon. Machine rebooted by operators. Services ok.

  • 2007-06-26: lxb2173 removed from production (LB). Will be reinstalled with the latest 3.1 middleware and moved in new cluster gridwms (new name: lb101).

  • 2007-06-26: rb117 removed from production (WMS for LHCb). Will be reinstalled with the latest 3.1 middleware and moved in new cluster gridwms (new name: wms101).

  • 2007-06-26: One CMS user submitted jobs to the CMS RBs (rb107, rb119 and rb122) with two anyMatch clauses in the JDL file, causing the workload manager to crash gain and again. Fixed by modifying one of the anyMatch clause in the input.fl file.

  • 2007-06-25: Vobox lxn1179 dedicated to Atlas removed from production. GOCDB updated.

  • 2007-06-21: Middleware upgrade (update 26 for gLite 3.0). Nodes types involved: lcg-RBs.

  • 2007-06-20: no_contact alarm triggered on rb118 last night. Machine rebooted by operator. Some services have not been restarted automatically.

  • 2007-06-18: proxy_renewal service restarted on rb117. Machine back in production.

  • 2007-06-14: vm_kill alarm triggered on rb117. Following the procedure, this node has been rebooted. Developers contacted.

  • 2007-06-14: swap_full alarm triggered on rb117. Standard procedure applied by the operator.

  • 2007-06-13: Sandbox partition on rb102 full due to large sandbox output files generated by a CMS user. A script has been written to delete the content of these files automatically.

  • 2007-06-11: New UI lxn1194 dedicated to GD with external connectivity for the globus port range put in production today.

  • 2007-06-11: Installation of package lcg-fw on lxb2173 (gLite LB 3.1 used for tests) and rb201 (gLite LB 3.0). The new set defined in the firewall database is GD_LB.

  • 2007-06-08: CA update to version 1.14-1 on WMS and RB nodes.

  • 2007-06-08: monb002, monb003: running 'spma_wrapper.sh' in order to perform CA 1.14 upgrade

  • 2007-06-08: putting back rb108 to production (configured on monb002, monb003 for all VOs used there ('ops', 'dteam', 'cms', 'atlas', 'alice')

  • 2007-06-06: putting back rb108 to production (configured on monb002, monb003 and on the development SAM machine lxn1182 for the 'ops' VO

  • 2007-06-06: Middleware upgrade on rb126 (gLite WMS 3.1 nodes).

  • 2007-06-06: Allow access to the MySQL database on all the WMS nodes in production for the Real Time Monitoring (RTM). A new set name IT CC GRID WMSRB RTM has been created for that purpose.

  • 2007-05-31: rb107 completely stuck. Machine rebooted by sysadmins and back in production.

  • 2007-05-30: Middleware upgrade (update 25 for gLite 3.0). Nodes types involved: lcg-RBs.

  • 2007-05-25: Update of files /opt/glite/etc/glite_wms_wmproxy.gacl and /opt/glite/etc/vomses because of some problems with job submission through the wmproxy on the WMS nodes.

  • 2007-05-23: Sandbox directory full once again on rb102.

  • 2007-05-22: Problem with the glite account on all the WMS nodes which had the bad gid. The only way to have clean machines was to reboot them all. It fixed the problem.

  • 2007-05-22: The alarm nospma_present has been disabled on all the gLite WMS 3.1 nodes.

  • 2007-05-22: Upgrade of package lcg-vomscerts to version 4.5.0-1 on all the WMS 3.1 nodes (part of update 24 for gLite 3.0).

  • 2007-05-22: Middleware upgrade (update 24 for gLite 3.0). Nodes types involved: WMS 3.0 and LB 3.0.

  • 2007-05-21: Load-balanced DNS alias lhcb-wms.cern.ch created and applied to the WMS 3.1 nodes dedicated to LHCb (rb112 and rb117).

  • 2007-05-21: Sandbox partition full on rb102. One user created huge files (>20GB).

  • 2007-05-18: Middleware upgrade (update 24 for gLite 3.0). Nodes types involved: lcg-RBs.

  • 2007-05-14: Installation of package CERN-CC-settodesiredstate-DM on all LCG RBs nodes and WMS nodes. This package contains two scripts (/usr/libexec/SetToDesiredState.lcgrb and /usr/libexec/SetToDesiredState.gridrb) used to disable alarms and actuators when a machine is put in maintenance. All these alarms and actuators are restored when the node is back in production.

  • 2007-05-11: Installation of package gridview-wsc-js package (GridView) on all LCG RBs nodes.

  • 2007-05-10: SAM WMS node rb108 removed from production because of some problems. Actually testing what is wrong. All jobs in state "Hold" removed from the condor queue.

  • 2007-05-08: minor upgrade of the gLite middleware on rb111 and rb117 (gLite WMS 3.1 dedicated to LHCb).

  • 2007-05-08: WMS nodes dedicated to Alice (rb111 and rb116) upgraded to version 3.1.

  • 2007-05-08: kickstart files SL3-SEC-* and SL4-SEC-* support ntp installation and configuration, and the FQDN of the host is now included in files /etc/hosts and /etc/sysconfig/network (bug in anaconda).

  • 2007-05-08: root access restriction on lxdpm101 (SAM DPM).

  • 2007-05-07: package lcg-fw installed on the DPM nodes in production (cluster griddpm).

  • 2007-05-06: Upgrade of the gridview-wsc-js package (GridView) to version 1.1.2-1 on rb104 and rb124.

  • 2007-05-04: Installation of a new package related to gridview (gridview-wsc-js-1.1.1-1) on rb104 and rb124 (templates profile_rb104.tpl and profile_rb124.tpl). Note also that the package perl-SOAP-Lite has been upgraded to version 0.69-1_cern on the whole cluster lcgrb (cf template pro_software_lcgrb_slc3.tpl).

  • 2007-05-03: jobs submission blocked on rb101 due to the huge size of file input.fl. Services NS, WMproxy and LM have been stopped.

  • 2007-05-02: add 90 dteam pool accounts on lxdpm101.

  • 2007-04-30: root access restriction on nodes belonging to clusters lcgrb and gridrb.

  • 2007-04-27: A lot of tickets for WM_WRONG alarms on rb107, rb119 and rb122 machines (LCG RBs dedicated to CMS). This is due to a bug in the workload manager if the JDL used contains several anyMatch clauses (see bug #21973). Fixed in the morning.

  • 2007-04-19: Middleware upgrade (update 21 for gLite 3.0). Nodes types involved: WMSLB 3.0 and lcg-RB.

  • 2007-04-19: no_contact alarm triggered two times this morning. Machine back in production but we don't know the reason of the crash (will be investigated by the sysadmin?).

  • 2007-04-18: renew host certificates on rb101 to rb108. Services restarted.

  • 2007-04-18: package lcg-fw upgraded to version 1.5 on all the LCG RBs in production (CDB template pro_software_lcgrb_slc3.tpl updated).

  • 2007-04-10: /etc/cron.d/glite-wms-check-daemons.cron error output redirected to /dev/null on all WMS 3.0 and WMS 3.1 nodes. Added comment on the bug #21909.

  • 2007-04-05: Worload manager on rb107, rb119 and rb122 is dying all the time due to the bug #20973. The responsible user has been informed. The problem has been fixed.

  • 2007-04-03: Middleware upgrade (update 20 for gLite 3.0). Nodes types involved: WMSLB 3.0 and lcg-RB.

  • 2007-04-02: all glite-WMS (3.0 and 3.1) nodes (rb102 rb103 rb108 rb109 rb111 rb116 rb118)
    • Remove backup copy of /etc/cron.d/glite-wms-purger.cron because it is run also. That can create high I/O load.

  • 2007-04-02: Due to the change of 3 switches in the computer center, all the services on the nodes rb101 to rb108 have been stopped during the morning (see here).

  • 2007-03-27: rb107
    • OS hung. Restarted. High load. Condor very slow. One user was submiting too many jobs. Back to normal after some time.

  • 2007-03-26: alarm NOSPMA has been disabled on all the gLite WMS 3.1 nodes.

  • 2007-03-23: monb002, monb003
    • Previous operation had to be rolled back because of a problem with the RPMs. It will be carried out on Monday 26th.

  • 2007-03-23: monb002, monb003
    • GFAL-client-1.9.0-1.i386.rpm, lcg_util-1.5.0-1.i386.rpm packages installed (using quattor). The packages are certified, but not in production yet. SAM tests urgnetly needed the fix provided in these packages.

  • 2007-03-23: Some files added by the Gridview team on rb104 and rb124:
    • Created a dir /opt/gridview
    • All the files and directory inside /opt/gridview belongs to us.
    • Any log and other things will be done in /opt/gridview.
    • Necessary permissions are given to files and dir in /opt/gridview.

  • 2007-03-23: All the gdrbxx nodes removed from production definitely.

  • 2007-03-22: rb127 was down last night (around 02:15) and has been rebooted by the sysadmins (around 02:35). All the services are up and running.

  • 2007-03-22: Middleware upgrade (update 18 for gLite 3.0). Nodes types involved: WMSLB. The /opt/glite/etc/glite_wms_wmproxy.gacl file has been overwritten with this upgrade... Some ggus ticket assigned to CERN-PROD related to this problem. It has been fixed.

  • 2007-03-21: monb002, monb003: previous entry undone (the request was due to a misunderstanding)

  • 2007-03-20: httpd installation for 'alice' VO VOBOX tests on monb002, monb003
    • httpd installed by adding include pro_service_http; in their quattor profiles
    • /var/www/html/alice directory created with owner samalice:c3

  • 2007-03-13: Interactive access and root access granted to some gridview developers on rb104 and rb124.

  • 2007-03-13: Middleware upgrade (update 17 for gLite 3.0). Nodes types involved: WMSLB and LB.

  • 2007-03-13: Upgrade to gLite 3.1 of rb112 (WMS for Atlas).

  • 2007-03-12: rb110 (WMS dedicated to Atlas) is configured to use a separate LB (rb201).

  • 2007-03-12: Reinstallation from scratch of nodes rb112 and rb117 with gLite WMS 3.1.

  • 2007-03-12: About 20 new lemon exceptions configured on all the LCG RB nodes. There is a wiki page available here. The operational procedure guide has been updated accordingly.

  • 2007-03-10: Alarm raid_tw triggered on rb115. Fixed by sysadmins.

  • 2007-03-09: Middleware upgrade (update 16 for gLite 3.0). Nodes types involved: UIs, VOBox, RBs, classic SE and WMS 3.0 nodes.

  • 2007-03-05: UI for CMS lxb1930 removed definitively from production.

  • 2007-03-05: Configuration of the SAM BDIIs bdii109 and bdii110 changed. Value of variable BDII_UPDATE_URL in /opt/bdii/etc/bdii.conf modified to:
                BDII_UPDATE_URL=http://goc.grid-support.ac.uk/gridsite/bdii/BDII/www/bdii-sam.conf

  • 2007-03-05: Middleware upgrade (update 15 for gLite 3.0). Nodes types involved: WMSLB and LB.

  • 2007-03-05: UI for CMS lxb1930 to remove definitively from production

  • 2007-03-01: Upgrade manually package edg-fabricMonitoring-agent on rb110, rb125, rb126 and rb201 (experimental nodes).

  • 2007-03-01: Package eela-vomscerts upgraded on all the gLite WMS 3.0 and LCG RBs.

  • 2007-02-28: smart_selftest alarm triggered on monb003. Fixed by sysadmins.

  • 2007-02-28: UI for LHCb lxb2008 removed definitively from production.

  • 2007-02-26: New configuration for the SAM cron jobs on monb002 and monb003. VO CMS is now supported.

  • 2007-02-26: UIs for Atlas lxb0725 and lxb0726 removed definitely from production.

  • 2007-02-26: New LCG RB rb127 for SAM put in production.

  • 2007-02-23: Job submission blocked on gdrb01 and gdrb03.

  • 2007-02-22: VO compass supported on rb104 and rb124.

  • 2007-02-20: Job submission blocked on gdrb02, gdrb04 to gdrb11.

  • 2007-02-20: Middleware upgrade (update 14 for gLite 3.0). Nodes types involved: classic SE, LCG RBs, gLite WMS, UIs and VOBOX.

  • 2007-02-20: rb115 rebooted during the night (no_contact alarm, black screen). Back in production.

  • 2007-02-14: New experimental gLite WMS 3.1 rb125 put in production.

  • 2007-02-12: CA update to version 1.12-1. Patch #998 applied on rb104 (the config_condor script from yaim has not been executed yet).

  • 2007-02-12: New LCG RBs rb119 to rb124 put in production.

  • 2007-02-11: rb109 down during one hour... Machine rebooted and services restarted. There was a FANCOUNT_WRONG alarm for this machine after the reboot which has been fixed in the evening. To be checked by sysadmins.

  • 2007-02-12: Alarm smart_selftest on monb003 (SAM UI). To be checked by sysadmins.

  • 2007-02-10: Alarm HWSCAN_WRONG on bdii110 (SAM BDII). To be checked by sysadmins.

  • 2007-02-09: Installation of package lcg-fw on monb002 to monb005. Registration of these machines in the firewall database (cluster GD_UI and GD_SAM).

  • 2007-02-09: Alarm RAID_TW triggered on rb118 (disk /dev/sda1). Fixed in the afternoon.

  • 2007-02-09: Configuration of UI lxb1930 to submit jobs on gdrb08, gdrb06 and rb107 (LCG RBs), and rb102 and rb109 (gLite WMS). See files /opt/edg/etc/cms/edg_wl_ui.conf and /opt/glite/etc/cms/glite_wmsui.conf.

  • 2007-02-09: 6 new LCG RBs assigned to experiments:
    • rb119: CMS
    • rb120: Alice
    • rb121: Atlas
    • rb122: CMS
    • rb123: LHCb
    • rb124: All VOs.

  • 2007-02-08: "hole" in the cpu utilization of the three BDIIs bdii105, bdii108 and bdii112 between 17h00 and 18h00.

  • 2007-02-08: Middleware upgrade (update 13 for gLite 3.0) on rb113 and rb106. The new version of Condor condor-lcgrb-1.0.0-3.i386.rpm has been installed. Machines still remaining: rb104.

  • 2007-02-07: Middleware upgrade (update 13 for gLite 3.0). Nodes types involved: classic SE, LCG RBs, UIs and VOBOX. Note that on some LCG RBs, the patch #968 (corresponding to package condor-lcgrb-1.0.0-2.i386.rpm) had already been installed and there is a new version of this package (condor-lcgrb-1.0.0-3.i386.rpm corresponding to patch #998) available in update 13. Actually:
    • rb106 has not been upgraded yet (no condor-lcgrb-1.0.0-3.i386.rpm installed). Mail sent to Dietrich Liko.
    • rb113 and rb104 have still the old version condor-lcgrb-1.0.0-2.i386.rpm installed.
    • rb114, rb105, rb107 (this morning) and rb115 (this morning) have the new version condor-lcgrb-1.0.0-3.i386.rpm installed.

  • 2007-02-06: Stopping/restarting edg-wl-proxyrenewal and edg-wl-ns services on gdrb04 (high system cpu utilization due to a broken pipe in a proxyrenewal process).

  • 2007-02-04: Patch #998 applied on rb114 and rb105 to make the condor-G components more robust (RPM taken from the PPS repository). Machine back in production.

  • 2007-02-05: gdrb06 down this morning. Nothing found in the logs. Machine back in production.

  • 2007-02-04: Patch #968 applied on rb105 to make the condor-G components more robust (RPM taken from the PPS repository). Machine back in production.

  • 2007-02-03: Alarm tmp_full triggered on rb118. Fixed by removing 2 huge Condor files.

  • 2007-02-01: gdrb04 down in the evening. Nothing found in the log files. Back in production.

  • 2007-02-01: Creation of a virtual cluster for SAM framework. The Lemon page for this cluster can be found here.

  • 2007-01-31: VO CMS supported again on lxn1183.

  • 2007-01-31: Alarm tmp_full triggered on rb108. Fixed by removing 3 huge Condor files.

  • 2007-01-30: bdii110 back in production this afternoon.

  • 2007-01-30: Jobs submission blocked on rb114 this morning, on rb106 this afternoon.

  • 2007-01-29: File /opt/glite/etc/glite_wms_wmproxy.gacl were overwritten on rb102, rb103, rb109 and rb110 in last update, changed them back. And also deleted the entries for other VOs there on rb102 and rb109 since these two machines are reserved for CMS.

  • 2007-01-29: Jobs submission blocked on rb105 and rb107 at the end of the afternoon. The new Condor rpm will be deployed on both nodes in the next few days.

  • 2007-01-29: Jobs submission allowed again on gdrb03.

  • 2007-01-28: Problem with one of the RAID disks on bdii110. Vendor call requested.
                [root@bdii110 root]# smartctl -l selftest  --device=3ware,1  /dev/twe0
                smartctl version 5.33 [i686-redhat-linux-gnu] Copyright (C) 2002-4 Bruce Allen
                Home page is http://smartmontools.sourceforge.net/
                === START OF READ SMART DATA SECTION ===
                SMART Self-test log structure revision number 1
                Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
                # 1  Short offline       Completed without error       00%       773         -
                # 2  Extended offline    Completed: read failure       90%       748         86833
                # 3  Short offline       Completed without error       00%       725         -
                # 4  Short offline       Completed without error       00%       701         -

  • 2007-01-27: Problem with one of the RAID disks on rb115. Fixed.

  • 2007-01-26: VO CMS supported temporally on rb104 in order to fix a problem with Condor on rb107 (Patch #968 will be applied next week on rb107 to make the condor-G components more robust).

  • 2007-01-26: Middleware upgrade (update 12 for gLite 3.0) on all the WMSLB in production (new CDB template pro_software_glite_2_4_8-0_glite-wmslb.tpl created and update of pro_software_packages_cern_slc3_glite3_0_wmslb.tpl).

  • 2007-01-26: Sandbox directory on gdrb03 full. New job submissions blocked. The main job submitter has been contacted and he will try to retrieve the job output of his files. Log monitor in a bad shape now due to some inconsistancy with the MySQL database. Maarten will try to restore the service. Fixed.

  • 2007-01-25: False alarm fancount_wrong on rb104 to rb107. Fixed.

  • 2007-01-25: Some services were not restarted automatically on all the WMS nodes. Fixed by hand.

  • 2007-01-25: Service edg_wl_lm was not restarted automatically on some LCG RBs nodes. Fixed by hand.

  • 2007-01-25: Need to reinstall the edg-fabricMonitoring-agent package on the LCG RBs nodes because the monitoring was not running anymore. I did the following things:
                rpm -Uvh --force edg-fabricMonitoring-agent-2.13.0-3.i386.rpm
                /etc/init.d/edg-fmon-agent restart

  • 2007-01-25: Major power cut at Cern. All the nodes were down. The reboot of the machines did not restart some services automatically, especially on the WMS and the RBs nodes. Moreover, two problems were discovered why boxes (eg. lxb0725, lxb0726 and monb00[2-5], rb201) were not able to start automatically.
    • DHCP clients tried to contact the DHCP server but the one was down at that moment. Network service picked up the old IP, then tried to contact (ping) gateway but it was not accessible also. As a result network services switched off the network interface. This shows the necessity for the ordered boot of CC services.
    • Problem with PXE boot. Requires manual intervention during booting process to switch off the graphical boot.

  • 2007-01-24: Modification of the cron job /etc/cron.daily/slocate.cron on rb101 to rb118 to exclude the directories /data01, /data02 and /data03 of the directories tree to check.

  • 2007-01-24: Modification of the cron job /etc/cron.daily/slocate.cron on voatlas01 and volhcb01 to exclude the directory /storage of the directories tree to check.

  • 2007-01-23: Middleware upgrade (update 12 for gLite 3.0) on all the machines in production (new CDB template pro_software_glite_2_2_7-0_glite-lb.tpl created for rb201 and update of pro_software_gridrb_lb_slc3.tpl; need to remove et put back the /etc/nospma file on this node to avoid automatic upgrade). The WMS have not been upgraded due to a problem with the packages included in this update.

  • 2007-01-19: Alarm tmp_full triggered on rb108. Need to delete the content of some condor files. Fixed.

  • 2007-01-19: Patch #968 applied on rb115 to make the condor-G components more robust (RPM taken from the PPS repository).

  • 2007-01-18: Patch #968 applied on rb113 to make the condor-G components more robust (RPM taken from the PPS repository).

  • 2007-01-17: monb002 and monb003 are using rb104 for testing the patch applied on it

  • 2007-01-12: Patch #968 applied on rb104 to make the condor-G components more robust (RPM taken from the PPS repository).

  • 2007-01-12: Kernel upgrade on rb101.

  • 2007-01-15: New gLite WMS rb118 put in production for SAM jobs submissions. Machine registered on the MyProxy server.

  • 2007-01-12: The faulty disk on monb003 has been rebuilt successfully by the sysadmins. Machine back in production.

  • 2007-01-12: CAs upgraded to version 1.11-1 on all machines in production.

  • 2007-01-12: directory tmp almost full on lxb7283. Condor files deleted to fix the problem.

  • 2007-01-12: Kernel upgrade on rb102.

  • 2007-01-11: Problem with one of the RAID disks on monb003. This disk is actually in degraded mode.

  • 2007-01-12: Add script /etc/cron.hourly/lcg-mon-job-status.cron in order to avoid to restart the service lcg-mon-job-status manually after a crash.

  • 2007-01-11: cron job files used for SAM job submissions on monb002 and monb003 created and configured in Quattor (see template files profile_monb002.tpl and profile_monb003.tpl).

  • 2007-01-09: In order to fix some problems on the gLite WMS machines rb1xx, patch #944 has been applied on all the nodes. The following packages have been installed via quattor (see CDB files pro_software_packages_cern_slc3_glite3_0_wmslb.tpl and pro_software_glite_2_4_7-0_wmslb.tpl):
                glite-lb-common-3.0.6-1.i386.rpm
                glite-lb-server-bones-2.1.5-1.i386.rpm
                glite-lb-server-1.3.9-1.i386.rpm
                glite-lb-logger-1.2.3-1.i386.rpm
                glite-lb-proxy-1.2.8-0.i386.rpm
                glite-security-gsoap-plugin-1.2.5-0.i386.rpm 

  • 2007-01-09: Package lcg-vomscerts-4.3.0-1 upgraded on all the gdrbxx nodes. The problem is that the middleware installed on these node is version 2.7.0, and the new certificate upgrades are not put in the old repository. It caused some problems when we try to get the status of some jobs on these RBs.

  • 2007-01-08: Problem fixed on the gdrbxx nodes concerning the ncm_cdispd_wrong alarm by disabling the relevant metrics (ncm-cdispd, exception.ncm_cdispd_wrong) for those nodes in CDB profile files (profile_gdrb01.tpl to profile_gdrb11.tpl):
                "/system/monitoring/metric/_5124/active" = false;  
                "/system/monitoring/exception/_30068/active" = false;

  • 2007-01-06: Remedy ticket CT392757 concerning rb114: problem with the fans (no_contact alarm). This machine is still in production. I need to ask to LHCb (Roberto Santinelli) before to remove this machine from production.

  • 2007-01-05: Alarm ncm_cdispd_wrong on all the gdrbxx nodes. This problem has been fixed by executing ccm-fetch and ncm-ncd --configure fmonagent on all the gdrbxx nodes. No it is not enough frown I sent an email to the experts...

  • 2007-01-05: Alarm tmp_full on gdrb06. There were indeed 3 huge file in directory /tmp. I deleted the content of these 3 files without removing the files themselves. Fixed.

  • 2007-01-05: Alarm extfswarning on gdrb06 due to some RAID disk error discovered by Maarten last night. All the services have been stopped (gridftp services excepted) and the machine has been put in maintenance. Machine rebooted and the alarm disappeared. However the MySQL database is corrupted. Trying to fix it.

  • 2007-01-05: Need to restart manually services edg-wl-jc, edg-wl-lm and edg-fmon-server on rb114. Fixed. This machine has been rebooted (no_contact alarm). All the services are back again.

  • 2007-01-03: Need to reinstall the edg-fabricMonitoring-agent package on rb113 because the monitoring was not running anymore. This monitoring is back again and the LAS ticket is closed. Note that this machine was down and has been restarted by a sysadmin. He didn't notice anything in the logs.

  • 2007-01-03: Remove lcg-registrar.cern.ch from GRIDMAP_AUTH variable in site-info.def and /opt/edg/etc/edg-mkgridmap.conf files.


  • 2006-12-31: gdrb06 back in production since the alarm disappeared.

  • 2006-12-26: Unrecoverable error on gdrb06 (problem with RAID disk /dev/hde). This machine has been put in maintenance:
                Dec 26 18:53:09 gdrb06 kernel: hde: drive_cmd: status=0x61 { DriveReady DeviceFault Error }
                Dec 26 18:53:09 gdrb06 kernel: hde: drive_cmd: status=0x61 { DriveReady DeviceFault Error }
                Dec 26 18:53:09 gdrb06 kernel: hde: drive_cmd: error=0x04 { DriveStatusError }
                Dec 26 18:53:09 gdrb06 kernel: hde: drive_cmd: error=0x04 { DriveStatusError }

  • 2006-12-24: Upgraded rb101 for patches 900 and 944 asked by ATLAS.

  • 2006-12-22: Grant access to the MySQL Database on all the new LCG RBs in production (rb104 to rb107, rb113 to rb115).

  • 2006-12-22: Reinstallation from scratch of rb104. This LCG RB now supports the following VOs: unosat, ops, dteam, sixt, gear, geant4.

  • 2006-12-21: VOMS configuration update on all the machines fully managed by Quattor (rbxxx nodes, voatlas01 and volhcb01).

  • 2006-12-20: Reinstallation from scratch of rb108. This machine becomes a gLite WMSLB for VO ops (and dteam) and has therefore been moved to cluster gridrb. This machine has been put in production at the end of the afternoon.

  • 2006-12-20: Middleware upgrade (update 11 for gLite 3.0) on all the machines in production.

  • 2006-12-20: Job submission blocked on rb104 this morning. This node will be reinstalled next Friday.

  • 2006-12-19: monb002 and monb003 are using rb110 as gLite WMS for VO OPS.

  • 2006-12-19: Problem with lxn1179. This machine is getting out of memory (kscand process running for a while). Simone Campana contacted. Problem fixed by stopping DQ2 site services on lxn1179.

  • 2006-12-19: Due to crash on rb114 last night, the MySQL database were corrupted (tables events and states had some errors). This has been fixed by David.

  • 2006-12-19: NO_CONTACT exception triggered on rb114 this morning. This machine has been rebooted by the operators, but there is a problem with the edg-fmon-agent service. Under investigation.

  • 2006-12-19: SMART_SELFTEST exception triggered on bdii110. There is a read failure problem on disk /c0/p1. Long smart test launched. Problem fixed by the sysadmins. To check the disks, the following commands should be executed (see /etc/smartd.conf file):
                smartctl -l selftest  --device=3ware,0  /dev/twe0
                smartctl -l selftest  --device=3ware,1  /dev/twe0

  • 2006-12-18: Reconfiguration of VO compass on gdrb01 and gdrb03. The host certificate of the VOMS server for this VO can be found in /etc/grid-security/vomsdir/dgrid-voms.fzk.de. New parameters are:
GRIDMAP_AUTH="ldap://lcg-registrar.cern.ch/ou=users,o=registrar,dc=lcg,dc=org  ldap://gridldap1.fzk.de/ou=People,ou=compass,dc=gridka,dc=de" 
VO_COMPASS_SW_DIR=$VO_SW_DIR/compass
VO_COMPASS_DEFAULT_SE=$CLASSIC_HOST
VO_COMPASS_STORAGE_DIR=$CLASSIC_STORAGE_DIR/SE01/compass
VO_COMPASS_QUEUES="compass"
VO_COMPASS_USERS=ldap://gridldap1.fzk.de/ou=compass,dc=gridka,dc=de
VO_COMPASS_VOMS_SERVERS="vomss://voms.fzk.de:8443/voms/compass?/compass/"
VO_COMPASS_VOMSES="compass dgrid-voms.fzk.de 15010 /O=GermanGrid/OU=FZK/CN=host/dgrid-voms.fzk.de compass" 

  • 2006-12-18: Reinstallation from scratch of rb105. This LCG RB is dedicated to VO Alice now.

  • 2006-12-16: Package gd-auth upgraded to version 1.1 on all the nodes in production and not managed by Quattor.

  • 2006-12-15: The configuration of the two UIs monb002 and monb003 has been changed:
    • In file /opt/edg/etc/edg_wl_ui_cmd_var.conf, line beginning by LoggingDestination has been commented out.
    • In file /opt/edg/etc/${VO}/edg_wl_ui.conf (where VO can be alice, atlas, cms, dteam and ops), lines beginning by NSAddresses and LBAddresses have been replaced on monb002 and monb003 respectively by:
NSAddresses = {"rb113.cern.ch:7772","rb115.cern.ch:7772"};       (resp. NSAddresses = {"rb115.cern.ch:7772","rb113.cern.ch:7772"};
LBAddresses = {{"rb113.cern.ch:9000"},{"rb115.cern.ch:9000"}};   (resp. LBAddresses = {{"rb115.cern.ch:9000"},{"rb113.cern.ch:9000"}};

  • 2006-12-15: No new job submissions are now allowed on rb105. This machine will be reinstalled from scratch next Wednesday 13 december.

  • 2006-12-14: gLite WMS rb116 and rb117 put in production for VO Alice and LHCb respectively.

  • 2006-12-14: rb115 put in production for SAM job submission. monb003 has been reconfigured to submit jobs on rb115.

  • 2006-12-13: Add rb115 entry on bdii103 and bdii104 (site-level BDIIs).

  • 2006-12-13: Installation and configuration of the second LCG RB rb115 for SAM.

  • 2006-12-13: Reinstallation from scratch of rb106. This LCG RB is now dedicated to VO Atlas mainly.

  • 2006-12-13: Middleware upgrade (update 10 for gLite 3.0) on rb112, rb111, rb110, rb109, rb103, rb102 and rb101.

  • 2006-12-12: VO EELA supported on rb108 and on all the CEs at CERN. Note that a package named eela-vomscerts containing the host certificate of the VOMS server for this VO must be installed.

  • 2006-12-12: LB server rb201 put in production again after reinstallation and reconfiguration. This machine supports the following VOs: atlas, alice, cms, lhcb, gear, geant4, unosat, sixt, dteam and ops.

  • 2006-12-11: Update 10 for gLite 3.0 released. Machines involved are:
    • LCG RBs from 3.0.4-0 to 3.0.5-0: rb104 to rb108, rb113 and rb114.
    • Classic SE from 3.0.4-0 to 3.0.5-0: voatlas01, volhcb01 and lxn1183.
    • UI from 3.0.9-0 to 3.0.10-0: lxb0725, lxb0726, lxb1930 and lxb2007.
    • VOBOX from 3.0.10-0 to 3.0.11-0: lxn1179.

  • 2006-12-11: No job submissions are now allowed on rb106. This machine will be reinstalled from scratch next Wednesday 13 december.

  • 2006-12-10: bkserverd process in infinite loop on rb112. Service restarted and problem fixed.

  • 2006-12-08: Installation of the gd_auth package on lxb2004 and lxb2008.

  • 2006-12-07: Installation of the gd_auth package on all the gdrbxx nodes, lxb0725, lxb0726 and lxb1930.

  • 2006-12-07: lxb2008 down this morning. No particular message on the screen. Machine rebooted and back in production.

  • 2006-12-07: New alias sam-bdii which is now pointing to bdii109 and bdii110 (SAM BDIIs) in a load-balanced way. rb113 is now configured to use sam-bdii.

  • 2006-12-06: Reinstallation from scratch of rb107. This machine will be assigned to CMS.

  • 2006-12-05: Two BDIIs bdii109 and bdii110 put in production for SAM (alias with load-balancing: sam-bdii).

  • 2006-12-04: Jobs submission on rb107 blocked (see rule GD_RB_BLOCKED). This machine will be reinstalled in few days

  • 2006-12-02: rb111 put in production as gLite WMSLB for VO Alice.

  • 2006-12-01: rb112 put in production as gLite WMSLB for VO LHCb.

  • 2006-11-29: Package glite-wms-ism is upgraded to version 1.5.15-1 to fix the CE disappering bug on all gLite WMS's.

  • 2006-11-29: rb109 is back in production. According to sysadmins, the air flow is apparently working fine now.

  • 2006-11-28: gLite UI and SAM client installed on monb002. Need to have the same configuration on monb003.

  • 2006-11-28: Big Yvan's mistake on rb106 (machine rebooted by error).

  • 2006-11-27: reinstallation of rb201 (bad partition disk configuration).

  • 2006-11-27: Add rb113 and rb114 entries on bdii103 and bdii104 (site-level BDIIs)

  • 2006-11-27: rb113 put in production as a LCG RB for SAME and SFT.

  • 2006-11-27: rb114 put in production as a LCG RB for LHCb.

  • 2006-11-27: It seems that there is an airflow problem on rb109. This machine will be shutdown tomorrow Tuesday 28 november and a vendor call will be opened.

  • 2006-11-25: High temperature on the RAID controller detected on rb109 (some tickets opened for this case). I asked to the sysadmins team to check it.

  • 2006-11-24: Late in the evening, rb108 crashed once again. It was perhaps due to the high temperature on the battery of the RAID controller. Machine rebooted, and services back in production.

  • 2006-11-24: Need to restart service lcg-mon-job-status status on gdrb02, gdrb09, gdrb10, rb106 and rb108.

  • 2006-11-23: Change GLITE_WMS_QUERY_TIMEOUT from the default value, 300, to 480 in /etc/glite/profile.d/glite_setenv.(c)sh on rb101 (requested by ATLAS).

  • 2006-11-23: New machines assigned to GD:
    • rb201: new Logging and Bookkeeping node dedicated to experiments (Atlas and CMS).
    • rb111: gLite WMSLB dedicated to VO Alice.
    • rb112: gLite WMSLB dedicated to VO LHCb.
    • rb113: LCG RB dedicated to VO ops.
    • rb114: LCG RB dedicated to VO LHCb.

  • 2006-11-22: Add "ExpiryPeriod = 21600;" in the WM section of glite_wms.conf and "Dagmanloglevel = 5;" in the JC section of glite_wms.conf on rb101 (requested by ATLAS).

  • 2006-11-21: Add user gianelle for interactive access on rb104.

  • 2006-11-21: Package lcg-fw upgraded on gdrbxx nodes, on rb104 to rb108, lxb2003 and lxn1183

  • 2006-11-21: Package lcg-fw installed on lxb0725, lxb0726, lxb1930, lxb2004 and rlxb2008 (UIs for experiments).

  • 2006-11-20: Need to restart service lcg-mon-job-status on rb106, gdrb01 and gdrb06. Fixed.

  • 2006-11-15: Middleware upgrade on rb104 to rb108 for lcg-RB 3.0.3-2 to 3.0.4-0.

  • 2006-11-15: Middleware upgrade on lxb0725, lxb0726, lxb1930 and lxb2007 for glite-UI 3.0.8-0 to 3.0.9-0.

  • 2006-11-15: Middleware upgrade on lxn1179 for glite-UI 3.0.8-0 to 3.0.9-0, and glite-VOBOX 3.0.9-0 to 3.0.10-0.

  • 2006-11-15: Middleware upgrade on voatlas01, volhcb01 and lxn1183 for glite-SE_classic 3.0.4-0 to 3.0.5-0.

  • 2006-11-12: rb106 down for some unknown reason. The sysadmin made a reset using the Ctrl+e sequence keys. Machine back in service now.

  • 2006-11-10: rb108 back in production (memory module exchanged).

  • 2006-11-09: NO_CONTACT alarm on rb108. Black screen and unable to reboot it. VO gear supported by rb105.

  • 2006-11-09: IP re-numbering successfully done on lxb1930 this morning. Back in production.

  • 2006-11-09: Installation of a top-level EGEE.BDII lxb2005 used by gdrb02 for SAM. The bdii service has been stopped on gdrb02 because of timeout errors when gdrb02 was loaded.

  • 2006-11-06: lxn1182 blocked (out of memory message on the screen). Need to reboot this machine. Fixed.

  • 2006-11-04: Hard disk hda dead on lxb2001.

  • 2006-11-04: Problem on lxn1183 due to the IP re-numbering. The former IP address was in file etc/hosts. Fixed.

  • 2006-11-03: New gLite UI and top-level BDII lxb0728 in production dedictated to SRMv2 tests.

  • 2006-11-02: Castor client upgrade (2.1.1-1 to 2.1.1-4) on volhcb01 and lxb2004.

  • 2006-10-31: Castor client upgrade (2.1.1-1 to 2.1.1-4) on lxb1930, lxb2003, lxb2007 and lxb2008.

  • 2006-10-31: lxb7283 in an infinite loop (HIGH_LOAD alarm triggered) due to a bug in the gLite middleware (processes glite-lb-bkserverd and glite-lb-logd). All the related services have been restarted. Fixed.

  • 2006-10-31: Kernel upgrade on gdrbxx nodes, lxn1183, lxn1179, lxb0725 and lxb0726.

  • 2006-10-31: Castor client upgrade (2.1.1-1 to 2.1.1-4) on gdrbxx nodes, lxn1183, lxn1179.

  • 2006-10-31: IP re-numbering for some machine in production (gdrbxx, lxn11xx and lxb07xx) this morning.
CondorG was unable to restart on gdrbxx because of the wrong IP adress for each host in file /etc/hosts.

  • 2006-10-30: rb108 crashed. Need to reboot this machine. Fixed.

  • 2006-10-28: Kernel upgrade on lxb2003.

  • 2006-10-26: Kernel upgrade on lxb2004.

  • 2006-10-26: Cleaning of all the big files on rb102. Fixed by Di.

  • 2006-10-25: kernel upgrade on lxn1176 failed. Need to check.

  • 2006-10-25: same problem (than for rb106 yesterday evening) on gdrb08. Fixed,

  • 2006-10-24: Problem on rb106 due to a single job submission with JDL requirements that cannot be handled by the WM (see bug #20973 in Savannah. It was also the same problem on gdrb08 yesterday). Fixed.

  • 2006-10-24: Minor upgrade on the gLite UI (lxb0725, lxb0726, lxb1930) and the VOBOX lxn1179.

  • 2006-10-24: Change the value of the variable APTMAILTO found in file /etc/sysconfig/apt-autoupdate to RB.Support@cernNOSPAMPLEASE.ch on all the LCG RBs.

  • 2006-10-24: Minor middleware upgrade on all the LCG RBs (package glite-rgma-api-python 5.0.3-1 to 5.0.4-1).

  • 2006-10-23: Sandbox partition on rb102 full. Ask to CMS before to remove the big files.

  • 2006-10-23: rb110 is a new gLite WMS for CMS and is now in production.

  • 2006-10-23: Production gridfts cluster updated to latest gLite patch (773, 787, 801, 825, 852).

  • 2006-10-23: Problem on gdrb08 due to a single job submission with JDL requirements that cannot be handled by the WM (see bug #20973 in Savannah). Fixed by Maarten.

  • 2006-10-21: Deployement of a patch on the nodes in production to fix a vulnerability (Torque/OpenPBS local root privilege escalation vulnerability).

  • 2006-10-20: alarm spma_error on lxb7283 (problem with gridsite-shared and gridsite-apache packages). Fixed by updating CDB.

  • 2006-10-20: Major security incident: Torque/OpenPBS local root privilege escalation vulnerability. A lot of sites have been switched down during the week-end.

  • 2006-10-20: Installation and configuration of a top-level BDII (without FCR) on gdrb02. gdrb02 does not query lcg-bdii anymore.
                 [root@gdrb02 root]# cat /opt/bdii/etc/bdii.conf
                 ............
                 BDII_AUTO_MODIFY=no 
                 BDII_UPDATE_LDIF= 
                 ............
                 [root@gdrb02 root]#

  • 2006-10-19: CAs upgraded to version 1.10-1 on all nodes in production.

  • 2006-10-18: Alias myproxy-fts points now to a new machine prod-px-fts. Former myproxy-fts (lxb0728) has been removed from production.

  • 2006-10-18: rb109 back in production (cooling problem solved).

  • 2006-10-18: middleware upgrade on rb104 to rb108 (lcg-RB 3.0.3-1 to 3.0.3-2), and on the special UIs used by the experiments (gLite-UI 3.0.6-0 to 3.0.7-0)

  • 2006-10-16: Remove archivers lxn1190, lxn1191 and lxn1193 from production (service tomcat stopped and cron job /etc/cron.d/check-tomcat disable).

  • 2006-10-16: Set BOOTPROTO=dhcp in file /etc/sysconfig/network-scripts/ifcfg-eth{0,1} on several machines in production in order to prepare the IP renumbering planned from 2006-10-31 to 2006-11-15:
    • Date: 31/10/2006 from 08:00am to noon
      • lxb0725 (gliteUI for Atlas).
      • lxb0726 (glite UI for Atlas).
      • lxb0728 (myproxy for FTS - this machine will be replaced soon by a new mid-range server).
      • gdrb01 to gdrb11 (LCG RBs).
      • lxn1179: VOBOX for Atlas.
      • lxn1180: SAM server.
      • lxn1181: SAM server backup.
      • lxn1182: SAM client.
      • lxn1183: Classical SE.

    • Date: 08/11/2006 from 08:00am to noon
      • lxb1930 (gLite UI for CMS).
    • Date: 15/11/2006 from 08:00am to noon
      • lxb2003 (Classic SE for LHCb).
      • lxb2004 (UI for LHCb).
      • lxb2008 (UI for LHCb).

  • 2006-10-14: processes related to job controller in infinite loop on rb106. Need to restart service edg-wl-jc. Fixed.

  • 2006-10-13: change email adress for variable MAILTO in file /etc/cron.d/glite-wms-check-daemons.cron on rb101 to rb103. This value should be set to wms.support@cernNOSPAMPLEASE.ch for this type of nodes. Need to do it on rb109 as well.

  • 2006-10-13: problem with one of the RAID disk on rb104 fixed.

  • 2006-10-13: Minor upgrade of the middleware on RBs rb104 to rb109 (lcg-RB 3.0.3-0 to 3.0.3-1).

  • 2006-10-11: temperature too high on the RAID disks on rb109:
                   [root@rb109 root]# tw_cli info c0

                   Unit  UnitType  Status         %Cmpl  Stripe  Size(GB)  Cache  AVerify  IgnECC
                   ------------------------------------------------------------------------------
                   u0    RAID-1    OK             -      -       149.001   OFF    OFF      OFF
                   u1    RAID-1    OK             -      -       232.82    OFF    OFF      OFF
                   u2    RAID-1    OK             -      -       232.82    OFF    OFF      OFF
                   u3    RAID-1    OK             -      -       232.82    OFF    OFF      OFF

                   Port   Status           Unit   Size        Blocks        Serial
                   ---------------------------------------------------------------
                   p0     OK               u0     153.38 GB   321672960     VDBE1BTCE521MP
                   p1     OK               u1     232.88 GB   488397168     VDB41BT4DE3GZC
                   p2     OK               u1     232.88 GB   488397168     VDK41BT4DWV9DK
                   p3     OK               u2     232.88 GB   488397168     VDK41BT4DYP4ZK
                   p4     OK               u2     232.88 GB   488397168     VDK41BT4DX10JK
                   p5     OK               u3     232.88 GB   488397168     VDK41BT4DYYWUK
                   p6     OK               u3     232.88 GB   488397168     VDK41BT4DXNVRK
                   p7     OK               u0     232.88 GB   488397168     VDK41BT4DY0BRK

                   Name  OnlineState  BBUReady  Status    Volt     Temp     Hours  LastCapTest
                   ---------------------------------------------------------------------------
                   bbu   On           No        Fault     OK       TooHigh  0      xx-xxx-xxxx

  • 2006-10-09: Partition /tmp full (no more inodes available) on rb101 and rb103. Fixed.

  • 2006-10-08: Partition /tmp full (no more inodes available) on rb103. Fixed.

  • 2006-10-07: Partition /tmp full (no more inodes available) on rb103. Fixed by deleting all the empty files in this directory.

  • 2006-10-06: Reconfiguration of the RAID disks on rb101, rb103 and rb109 in order to increase the number of inodes available on the partitions.

  • 2006-10-06: Problem with one of the RAID disks on rb104:
                   [root@rb104 root]# smartctl -l selftest  --device=3ware,3  /dev/twa0
                   smartctl version 5.33 [i686-redhat-linux-gnu] Copyright (C) 2002-4 Bruce Allen
                   Home page is http://smartmontools.sourceforge.net/

                   === START OF READ SMART DATA SECTION ===
                   SMART Self-test log structure revision number 1
                   Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
                   # 1  Short offline       Completed: read failure       20%      4496         17041
                   # 2  Short offline       Completed: read failure       40%      4472         17041
                   # 3  Short offline       Completed: read failure       10%      4448         17041
                   # 4  Short offline       Completed: read failure       40%      4424         17041
                   # 5  Short offline       Completed: read failure       30%      4400         17041
                   # 6  Extended offline    Completed: read failure       90%      4375         17041
                   # 7  Short offline       Completed: read failure       10%      4352         17041
                   # 8  Short offline       Completed: read failure       10%      4328         17041
                   # 9  Short offline       Completed: read failure       10%      4304         17041
                   #10  Short offline       Completed: read failure       40%      4280         17041
                   #11  Short offline       Completed: read failure       40%      4256         17041
                   #12  Short offline       Completed: read failure       40%      4232         17041
                   #13  Extended offline    Completed: read failure       90%      4207         17041
                   #14  Short offline       Completed: read failure       30%      4184         17041
                   #15  Short offline       Completed: read failure       10%      4160         17041
                   #16  Short offline       Completed: read failure       40%      4136         17041
                   #17  Short offline       Completed: read failure       30%      4112         17041
                   #18  Short offline       Completed: read failure       40%      4087         17041
                   #19  Short offline       Completed: read failure       40%      4063         17041
                   #20  Extended offline    Completed: read failure       90%      4038         17041
                   #21  Short offline       Completed: read failure       20%      4015         17041

                   [root@rb104 root]#
A ticket has been opened for this case: http://cern.ch/helpdesk/problem/CT371446&email=sysadmin-team@cern.ch. The disk will be replaced in the next few days.

  • 2006-10-05: Pb with the worload manager and the proxy renewal services on gdrb01. Need to restart these services. Fixed.

  • 2006-10-05: Castor client upgraded from version 1.7.1.5-1 to version 2.1.1-1 on volhcb01 (specific LHCb configuration, do not need to do the same on voatlas01 according to Simone).

  • 2006-10-05: Reconfiguration of the RAID disks on rb102. There was not enough inodes in one of the partitions.

  • 2006-10-05: upgrade of the CASTOR packages on classic SE volhcb01.

  • 2006-10-03: manually upgrade of the CAs to version 1.9 on some machines in production. It has not been done automatically because of the change of the APT repository.

  • 2006-09-27: We have now many gdrbxx machines in degraded mode:
    • gdrb03 (failure on hdg)
    • gdrb04 (failure on hde)
    • gdrb06 (failure on hde)
    • gdrb08 (failure on hdg)
    • gdrb10 (failure on hdg).
    • gdrb05 (failures on hde and hdg) and not used anymore in production as a RB.

  • 2006-09-27: Software RAID-I on gdrb06 in degraded mode (problem on hde). The machine has been rebooted successfuly.

  • 2006-09-27: All the exceptions found on the gdrbxx nodes have been solved. It was a problem with the lemon packages installed. I did the following thing to solve this problem:
                   for x in `/usr/bin/seq -w 4 11`; do 
                   go gdrb$x "REPO=http://swrep.cern.ch/swrep/i386_slc3 ;
                   wget $REPO/lemon-host-check-1.1.0-7.noarch.rpm $REPO/lemon-sensor-exception-1.2.1-2.i386.rpm
                   $REPO/lemon-sensor-sure-1.0.1-2.noarch.rpm $REPO/lemon-sensor-fio-1.2-10.noarch.rpm ; 
                   rpm -Uvh lemon-sensor-fio-1.2-10.noarch.rpm lemon-sensor-sure-1.0.1-2.noarch.rpm 
                   lemon-sensor-exception-1.2.1-2.i386.rpm lemon-host-check-1.1.0-7.noarch.rpm ; 
                   ccm-fetch ; 
                   ncm-ncd --co fmonagent "; 
                   done

  • 2006-09-27: Fixed the problem with the smartd_wrong exception on gdrbxx nodes by putting the following content in file /etc/smartd.conf (this file was empty):
                   /dev/hda -d ata -a -I 194 -I 7 -s (S/../../[^6]/01|L/../../6/00)
                   /dev/hde -d ata -a -I 194 -I 7 -s (S/../../[^6]/01|L/../../6/00)
                   /dev/hdg -d ata -a -I 194 -I 7 -s (S/../../[^6]/01|L/../../6/00)

  • 2006-09-26: Same problem on rb103 than for rb101 this morning. Fixed.

  • 2006-09-26: Middleware upgrade on lxn1179 (VOBOX for Atlas). Current version is now 3.0.6.

  • 2006-09-26: There was a problem with the gLite Logging and Bookkeeping processes (infinite loop ?) on rb101. It was impossible to stop cleanly these processes, so I killed them by hand using kill -9. I then stopped and restarted all the services and the load on rb101 is now ok. Fixed.

  • 2006-09-25: rb108 highly overloaded due to LHCb.

  • 2006-09-25: Need to restart service edg-fmon-agent on gdrb06 and gdrb07. There are a lot of exceptions on the gdrbxx machines. Need to be fixed.

  • 2006-09-25: Manually added new CA repository and updated CAs on gdrbxx machines. (vvidic)

  • 2006-09-25: VO unosat configured and supported on gdrb03, gdrb09 and gdrb10.

  • 2006-09-22: Network server on rb102 dead. Need to restart it by hand. Fixed.

  • 2006-09-13: update of the CAs to version 1.9.

  • 2006-09-06: Security updates done on rb104 to rb108, and on voatlas01 and volhcb01.

  • 2006-08-31: False NO_CONTACT alarm on all the gdrbxx machines.

  • 2006-08-31: Add AFS access to user fprelz on rb107.

  • 2006-08-30: job controller restarted on rb106 (2 gaph processes in infinite loop). Fixed.

  • 2006-08-30: minor upgrade of the classic SE voatlas01 and volhcb01 (current version is now 3.0.3).

  • 2006-08-28: daemon ntpd stopped on lxb1930. Need to restart it and to set the runlevel information for this service (was in off state). Fixed.

  • 2006-08-28: The gLite startup script failed partially during the boot sequence on rb101. Need to stop and restart all services. Fixed.

  • 2006-08-27: Service edg-wl-wm stopped on gdrb01. Need to restart it. Fixed.

  • 2006-08-25: It is now possible to access to gdrb07, gdrb09, gdrb10 and gdrb11 via Kerberos. The problem was due to a misconfiguration of the Kerberos database.

  • 2006-08-24: End of the kernel upgrade for all the machines in production and managed by GD.

  • 2006-08-24: Kernel upgraded on gdrb01, gdrb03, gdrb06, gdrb08, gdrb09, gdrb10, lxb0725, lxb0726, lxb1930, lxn1183, rb105, rb106 and volhcb01.

  • 2006-08-23: services edg-wl-wm and edg-wl-proxyrenewal restarted on rb105. Fixed.

  • 2006-08-23: service rfiod stopped on gdrb01 to gdrb11.

  • 2006-08-22: Kernel upgraded on gdrb02, gdrb04, gdrb07, lxb2003, lxb2004, lxb2008, lxb7026, rb107 and volhcb01.

  • 2006-08-21: Migration of all files found on the raid disk sdb to sda (except directory /var/edgwl/SandboxDir/ which is still on sdb) on rb108. This machine also supports VO LHCb now. We would like to compare the performance of rb107 and rb108 because we suspect that the raid disk could be a bottleneck (raid misconfiguration).

  • 2006-08-21: Process edg-wl-wm (worload manager) not running on gdrb01. Need to restart it manually. Fixed.

  • 2006-08-20: Modify manually configuration file /etc/cron.daily/slocate.cron and /etc/updatedb.conf used by slocate and updatedb in order to avoid the indexation of files contained in rb-state directory. Need to check if these configuration files are replaced during the update of the package slocate.

  • 2006-08-17: False NO_CONTACT alarm on all the gdrbxx machines.

  • 2006-08-16: False NO_CONTACT alarm on all the gdrbxx machines.

  • 2006-08-16: Need to restart some services (edg-wl-wm and edg-wl-proxyrenewal) on rb105. Fixed.

  • 2006-08-16: Need to restart edg-fmon-agent on gdrb08. The lemon monitoring was off on this machine during 6 days. Fixed.

  • 2006-08-15: Machines myproxy-fts, gdrb11, rb108 and rb104 rebooted with the new kernel.

  • 2006-08-10: Beginning of the kernel upgrade (kernel 2.4.21-47.EL.cernsmp) on all the machines in production. Still need to reboot the machines. Ask to EIS when it is possible to do it.

  • 2006-08-09: jobs monitoring tool deployed on all the RBs machines (gdrb01 to gdrb11, and rb104 to rb108) in order to monitor the number of running and idle jobs in the condor queue. Results available at the following link (only from inside the Cern): http://lxb1524.cern.ch/plots.html.

  • 2006-08-09: file systems from the disk servers were not remounted on lxn1183 after the last reboot. Fixed by Maarten.

  • 2006-08-08: CAs upgraded to version 1.8-1 on all machines in production.

  • 2006-08-08: Remove package shell-compat on rb104 to rb108 because there is a conflict between packages shell-compat and cern-compat-locallinks). Fixed.

  • 2006-08-01: Some modifications made by Ulrich on the globus-gridftp startup script on volhcb01 (see mail sent by Ulrich on 2006-08-01):
    • changed startup level from 55 to 99.
    • IP tables setup added in the start up script to make sure that the number of requests accepted is limited.

  • 2006-07-31: Mirror broken on volhcb01. Fixed.

  • 2006-07-28: major power cut at CERN.

  • 2006-07-28: volhcb01 down. Need to reboot it. Fixed.

  • 2006-07-27: CERN-CIC site definitely closed. Resources have migrated to CERN-PROD.

  • 2006-07-27: WM was dead on gdrb01, input queue 4MB. WM restarted. (vvidic)

  • 2006-07-27: New BDIIs in production (bdii105 and bdii106) used by experiments (Freedom of Choice for Resources -FCR- running on them). The aliases used for these machines are: exp-bdii, atlas-bdii and prod-bdii-exp.

  • 2006-07-26: Need to reboot lxb2007. Fixed.

  • 2006-07-23: Some host certificates expired on several RBs (gdrb07 to gdrb11). These certificates have been replaced and we experienced some problems with the LM and the JC services due to CondorG. Fixed.

  • 2006-07-24: Wiki page GmodRoleDescription created with the definition and duties fo the GMOD.

  • 2006-07-20: rb102 fully managed by GD now.

  • 2006-07-19: lxn1183 back in production this morning.

  • 2006-07-18: lxb1133 (alias lfc-lhcb) put in maintenance. Need to check the connexions between this machine and the database.

  • 2006-07-18: Problem with a fuse in the CC. lxn1183 is down due to this problem.

  • 2006-07-18: VO unosat configured on lxn1183.

  • 2006-07-12: Need to reboot lxn1179 (VOBox for Atlas). Fixed.

  • 2006-07-11: Security incident. Need to remove a ssh public key on all machines in production.

  • 2006-07-11: Problem found with the log monitor service on gdrb03. This service opened a lot of CondorG files and exceeded the number of file descriptors dedicated to it (ie. 1024). Other RBs checked since this problem could occur to them as well. Fixed.

  • 2006-07-11: Unable to connect to gdrb11. Need to reboot this machine. Fixed

  • 2006-07-10: End of the CERN-CIC site (this site will be removed from GOC DB).

  • 2006-07-10: lxn1183 (Classic SE used by Atlas, Unosat and Geant4) moved from CERN-CIC to CERN-PROD site.

  • 2006-07-10: lxn1179 (VOBox for Atlas) moved from CERN-CIC to CERN-PROD site.

  • 2006-07-10: lxn1184 (CE) and lxb2001 (Monbox) removed from CERN-CIC site. Services stopped on lxn1184.

  • 2006-07-10: Need to reboot lxn1179 (VOBox for Atlas). Out of memory error message on the screen. Fixed.

  • 2006-07-04: http and https servers configured and running on voatlas01.

  • 2006-07-04: New classic SE voatlas01 (alias: atlas-logs) in production. This SE will be used by Atlas to store their log files.

  • 2006-06-29: http and https servers configured and running on volhcb01.

  • 2006-06-28: New classic SE volhcb01 (alias: lhcb-logs) in production. This SE will be used by LHCB to store their log files.

  • 2006-06-28: Since this morning the sustained load on the three new WMS (rb101 to rb103) is around 20. Need to stop (some processes have to be killed by hand with -9, especially glite-lb-bkserv and glite-lb-logd) and restart all the services on these nodes. One bug found (see ). Fixed.

  • 2006-06-27: UI lxplus configured to support the new WMS nodes rb101 to rb103.

  • 2006-06-27: rb101 to rb103 (gLite WMS) are now in production. The VOs supported are:
    • rb101 (alias wms-atlas): dteam, ops and atlas.
    • rb102 (alias rb-cms): dteam, ops and cms.
    • rb103 (alias rb-alice, rb-lhcb, rb-shared): dteam, ops, gear, unosat, sixt, na48 and geant4.

  • 2006-06-26: Problem with the configuration of the CEs ce101 and ce102. Special accounts xxxprd were not included in the /etc/security/limits.conf file, generating problem on the RBs which were unable to determine the exit status of the user's job (see for example ggus ticket #9743).

  • 2006-06-23: Inconsistency between two rpms on rb104 to rb108:
    • Package edg-fabricMonitoring-agent-2.12.1-1 coming from Quattor/Lemon.
    • Package edg-fabricMonitoring-2.5.4-4 (coming from the middleware, more precisely metapackage lcg-RB) which provides the client for Lemon and Gridice.

We fixed the situation by 1) Installing package edg-fabricMonitoring-2.5.4-4 via apt-get; 2) Removing package edg-fabricMonitoring-agent-2.12.1-1; 3) Reinstalling package edg-fabricMonitoring-agent-2.12.1-1.

In this way, we now avoid errors when we are doing an apt-get upgrade or an apt-get dist-upgrade. See EdgFabricMonitoringConflictWithRBs for more details. Fixed.

  • 2006-06-23: Upgrade of lcg-vomscerts-4.2.0-1 not done automatically on rb104 to rb108 (this problem has been discovered thanks to ggus ticket #9476). The auto-update of the middleware packages has not been done before because the package apt-autoupdate has not been installed by default by quattor. Fixed.

  • 2006-06-22: file server lxfsrk524 in bad shape this morning (services nfs and portmap down). I restarted these services. Need to reboot lxb2003, lxb2004, lxb2008 and lxn1183 because of the problem with lxfsrk524. Fixed.

  • 2006-06-22: Due to the power cut at Cern last night, the apt-autoupdate of the CAs has failed on all the machines in production managed by GD (host grid-deployment.web.cern.ch was unreachable). I updated these packages manually this morning. Fixed.

  • 2006-06-22: major power cut last night in the CC. All the machine in production down and restarted this morning. Some services restarted manually on the RBs (rb105, rb106, rb107). The other RBs (gdrbxx) have not been touched by this power cut.

  • 2006-06-19: 100 new pool account created on rb104 to rb107 for VO alice, atlas, cms and lhcb respectively. Pool accounts aliceprd, atlasprd, cmsprd and lhcbprd also created on the same machines.

  • 2006-06-19: new rpm CERN-CC-tmpwatch-1.3-2 installed on rb104 to rb108. The previous version of this rpm caused some problems on these RBs last week (see 2006-06-16).

  • 2006-06-17: lxb2003 frozen. It seems to be a problem with AFS (kernel panic on screen). Machine rebooted only on 2006-06-19 due to the week-end. Fixed now. Note that a new machine has been requested to FIO two weeks ago.

  • 2006-06-16: Problem detected on all the RBs (rb104 to rb108) due to the bug in the CERN version of the "tmpwatch" system rpm (see 2006-06-13). A backlog of some 8000 logging events had built up (in /var_tmp on rb107) and was only getting processed very slowly, because there was competition from a continuous stream of new job submissions. Fixed by Maarten.

  • 2006-06-14: VO ops configured on rb104 to rb108. Account opssgm has been also created.

  • 2006-06-13: Problem on rb104 to rb108. Symbolic link /var/tmp (which points to /rb-state/var/tmp) disapeared and has been replaced by regular file /var/tmp (due to cron job /etc/cron.hourly/tmpwatch.sh). Fixed (patch proposed by Maarten).

  • 2006-06-12: FTS service stopped for DB and Castor upgrade. Resumed OK. (mccance)

  • 2006-06-06: FTA_WRONG SURE alarm installed on gridfts cluster (mccance)

  • 2006-06-06: lxn1183, lxb2005, lxb2007 and lxn1194 upgraded to gLite 3.0.0.

  • 2006-06-01: Swap full on gdrb03. Fixed.

  • 2006-06-01: lxb2009 (i.e. mon.cern.ch) removed from production. Alias mon.cern.ch will point to monb001 (FIO managed official monbox).

  • 2006-06-01: lxb0725 and lxb0726 upgraded to gLite 3.0.0.

  • 2006-05-31: Same persistent problem with lxb2003 due to a mis-configuration of the disk server lxfsrk524. This problem seems to be fixed now.

  • 2006-05-30: RBs rb104 to rb108 put in production. VOs supported are (VO ops not supported yet):
    • rb104 (alias rb-alice): dteam and alice.
    • rb105 (alias rb-atlas): dteam and atlas.
    • rb106 (alias rb-cms): dteam and cms.
    • rb107 (alias rb-lhcb): dteam and lhcb.
    • rb108 (alias multi-vo-rb): dteam, gear, unosat, sixt, na48 and geant4.

  • 2006-05-30: new myproxy server named prod-px with alias myproxy. This node is now managed by FIO. Former myproxy server lxn1192 will be retired.

  • 2006-05-29: Problem with lxb2003 again (out of memory messages due to gridftp connexions). Need to reboot it. Fixed.

  • 2006-05-26: Channel cleanup: Unused Tier-0 to Tier-1 / Tier-1 to Tier-0 / wildcard channels removed from gridfts (mccance)

  • 2006-05-26: Added ops VO to gridfts (mccance)

  • 2006-05-24: Created gLite environment on fts10[6-8] (mccance)

  • 2006-05-24: Problem with the rpm database on lxb2003. Need to regenerate this db. Fixed.

  • 2006-05-23: problem during the job submission on gdrb09. Should be fixed now.

  • 2006-05-23: Partition /var full on lxb2003. Fixed.

  • 2006-05-23: Need to start service edg-wl-proxyrenewal on gdrb03. This service was stopped. Fixed.

  • 2006-05-23: Scheduled intervention on FTS to change channel definitions to use GOCDB site names. (mccance)

  • 2006-05-22: Pb on gdrb03. Unable to submit jobs. Fixed.

  • 2006-05-20: Same problem on lxfsrk524. Fixed.

  • 2006-05-19: Unable to write (for normal users) on the disk server lxfsrk524 mounted from lxn1183. Need to reboot the disk server. Fixed.

  • 2006-05-19: DTEAM background transfers switched over to FTS validation cluster on fts00[1-5] for Oracle 10gR2 validation (mccance)

  • 2006-05-18: FTS history cleanup DBMS job installed on lcg_fts_prod account (mccance)

  • 2006-05-17: FTS GridView data collection trigger installed on lcg_fts_prod account (mccance)

  • 2006-05-17: All FTS channels switched Active again (mccance)

  • 2006-05-16: All FTS channels switched Inactive to avoid draining queue (Castor still recovering from power failure) (mccance)

  • 2006-05-16: update for edg-mkgridmap (2.6.1) on all the RBs (gdrb01 to gdrb11). This update fixes a problem which occurs when a VOMS or LDAP server is unavailable at the time the grid-mapfile is created (every 6 hours).

  • 2006-05-16: All LHCB jobs are failing registration to LFC since yesterday CERN power problem. Fixed ?

  • 2006-05-16: Unable to ping gdrb10 this morning. Bad reboot of the machine after EXTFSWARNING alarm. Fixed

  • 2006-05-16: power supply problem on lxn1181 (spare proxy). ITCM ticket generated.

  • 2006-05-16: major power cut at Cern. Need to check all the machines in production, especially the RBs.

  • 2006-05-16: FTS now publishing load-balanced alias in EGEE.BDII as prod-fts-ws.cern.ch (mccance)

  • 2006-05-16: Bad CAs on the machines in production (due to apt-autoupdate). Need to re-install them by hand to version 1.2-1.

  • 2006-05-16: Pb with raid disk /dev/hde on gdrb10 this morning. Machine rebooted and no more problem detected. MFixed.

  • 2006-05-15: FTS servers on production fts101, fts102, fts103, fts104, fts105 upgraded to 3.0 release. Pilot test cluster (fts001 - fts006) upgraded (mccance)

  • 2006-05-15: FTS servers (web-service nodes only) on production fts103 and fts104 upgraded to 3.0 release (mccance)

  • 2006-05-09: Partition dedicated to Atlas full on lxn1183 (size of partition: 1.8TB).

  • 2006-05-09: Problem with the two CE ce101 and ce102 due to a configuration error. Fixed this morning.

  • 2006-05-06: Problems on ce102 (VM_KILL, GRID_GRIS_WRONG, NO_ CONTACT, MIRROR_BROKEN). Will be fixed by FIO.

  • 2006-05-02: AFS error on gdrb10. Need to reboot this machine. Fixed.

  • 2006-04-24: bdii103 and bdii104: Final configuration of LFC - now using GRIS on LFC nodes to publish information (jamesc).

  • 2006-04-24: bdii103 and bdii104: entries removed for RLS since the nodes have been taken out of production by IT-PSS (jamesc).

  • 2006-04-24: /var filled up on lxb2003. Fixed and run rotate script for /var/log/messages (vvidic)

  • 2006-04-24: upgraded lcg-CA and lcg-yaim rpms on lxb2003 (vvidic)

  • 2006-04-20: centralized firewall configuration installed on myproxy, myproxy-fts, lxn1181 (spare proxy), lxn1178, lxb2009, lxn1190, lxn1191 and lxn1193.

  • 2006-04-17: gdrb08 blocked. Need to reboot it. Fixed.

  • 2006-04-14: ce102. blocked. Need to reboot it and to restart globus MDS. Fixed.

  • 2006-04-12: srm.cern.ch published by prod-bdii (i.e. bdii103 and bdii104). For this, file cern-cic-static.sh updated.

  • 2006-04-05: Updated lcgdm-mkgridmap.conf on all LFC nodes to be the same as the one generated by yaim. James

  • 2006-03-31: VO ops configured on all the RBs (gdrb01 to gdrb11).

  • 2006-03-30: lfc005 updated via Quattor to upgrade 4 RPMS: LFC-server-oracle, LFC-client, LFC-interfaces and lcg-dm-common to version 1.5.5-2.

  • 2006-03-30: VO ops configured on gdrb02, gdrb11, lxn1183 and lxn1184. Need to configure the other RBs.

  • 2006-03-30: kernel upgraded in lfc001 to version: 2.4.21-40.EL.cernsmp

  • 2006-03-30: kernel upgrade on all production nodes.

  • 2006-03-30: lfc001 updated via Quattor to upgrade 4 RPMS: LFC-server-oracle, LFC-client, LFC-interfaces and lcg-dm-common to version 1.5.5-2.

  • 2006-03-30: Move all services registered from CERN-SC to CERN-PROD (i.e. LFC nodes, myproxy-fts, castorgridsc and prod-fts-ws).

  • 2006-03-29: kernel upgrade on all RBs (gdrb01 to gdrb11).

  • 2006-03-29: update of the site BDIIs bdii103 and bdii104 to add the ops VO tho lfc-shared.cern.ch (updated by James).

  • 2006-03-29: update of the RBs gdrb01 to gdrb11 to patch #701.

  • 2006-03-27: alarm LFC_DB_ERROR triggered on lfc009.

  • 2006-03-27: High load (> 64 frown ) on lxb2003 due to a lot of connections via gridftp. Need to reboot it. Fixed.

  • 2006-03-23: CROND_WRONG alarm triggered on gdrb04. Need to kill some processes hanging related to cron daemon edg-mkgridmap. Fixed.

  • 2006-03-23: centralized firewall configuration installed on lxb2003.

  • 2006-03-23: VO ops configured on lfc-shared, lfc-dteam-test and lfc001 (file /opt/lcg/etc/lcgdm-mapfile modified).

  • 2006-03-22: after the reboot of lfc-shared (due to kernel upgrade), configuration for VO unosat disappeared. Files edg-mkgridmap.conf and lcgdm-mkgridmap updated by James and Maarten. Fixed.

  • 2006-03-22: service pbs_mon stopped on lxb2003 and lxn1183 (not needed for this type of nodes).

  • 2006-03-22: centralized firewall configuration installed on lxn1184 and lxn1183.

  • 2006-03-22: kernel upgrade done on all the LFC nodes (lfc001 to lfc011).

  • 2006-03-22: all the LFC nodes (lfc001 to lfc011) have been upgraded to LFC 1.5.4.

  • 2006-03-21: service mysql stopped on lxn1184 (not needed for this type of node).

  • 2006-03-20: Current version of LFC installed on the LFC nodes (latest version is LFC 1.5.4):
    • lfc001: LFC 1.4.1.
    • lfc002 (lfc-atlas-test): LFC 1.5.4.
    • lfc003 (lfc-cms-test): LFC 1.5.4.
    • lfc004 (lfc-atlas): LFC 1.4.1.
    • lfc005 (lfc-dteam-test): LFC 1.4.5.
    • lfc006 (lfc-shared or lfc-dteam): LFC 1.5.4.
    • lfc007 (lfc-alice): LFC 1.4.1.
    • lfc008 (lfc-atlas): LFC 1.4.1.
    • lfc009 (lfc-cms): LFC 1.4.1.
    • lfc010 (lfc-lhcb): LFC 1.4.1.
    • lfc011 (lfc-lhcb-ro): LFC 1.4.1.

  • 2006-03-17: Kernel needs to be upgraded on all the LFC nodes (lfc001 to lfc011). Planed for next week.

  • 2006-03-17: Upgrade of LFC on lfc-shared to version 1.5.4.

  • 2006-03-16: Misconfiguration of the SE name on lxn1184. File site-info.def modified and yaim reexecuted. Fixed.

  • 2006-03-15: Need to restart maui service on lxn1184 (Job submission via SFT failed for this reason). Fixed.

  • 2006-03-15: VO gear configured and now supported on gdrb01, gdrb03 and lxn1183.

  • 2006-03-14: Update of the edg-mkgridmap.conf on all nodes in the LFC cluster (lfc001 to lfc011) to use VOMS (they were just using LDAP and were missing new people).

  • 2006-03-13: New alias lhcbui created which points to lxb2004 (UI for LHCB experiments).

  • 2006-03-13: VO gear configured and now supported on lfc-shared.

  • 2006-03-11: lxb2008 installed as a new UI for LHCB experiments.

  • 2006-03-11: Problem with AFS on lxb2004 (machine blocked, kernel panic on the screen). Need to reboot it. Fixed.

  • 2006-03-10: Upgrade of LFC on lfc-atlas-test and lfc-cms-test to version 1.5.4.

  • 2006-03-08: Installation of new package lcg-mon-job-status-2.0.7-1_sl3.noarch.rpm on all the RBs in production (Patch #690).

  • 2006-03-08: Bug found in proxy configuration. This affects myproxy and myproxy-fts (cf. file /etc/init.d/myproxy-generate-config.pl). Fixed by Maarten.

  • 2006-03-07: installation of LCG 2.7.0 finished for all the production nodes.

  • 2006-03-06: VO Atlas supported on lxn1183 (lxfs5592 mounted on this machine).

  • 2006-03-06: Problem with the PBS server on lxn1184. Need to shutdown the gatekeeper, mds and bdii. These services have been restarted successfuly. Fixed.

  • 2006-03-06: LCG 2.7.0 installed on myproxy-fts and lxb2037. UI for Atlas lxb0725 and lxb0726 upgraded too.

  • 2006-03-03: Almost all the machines in production have been upgraded to LCG 2.7.0. Only some UIs for experiments (lxb0725, lxb0726, lxb1930, lxb2004 and lxb2037) and myproxy-fts have still LCG2.6.0 installed.

  • 2006-02-28: gdrb04, gdrb06, gdrb07 and gdrb08 upgraded to LCG 2.7.0.

  • 2006-02-28: The following nodes have been switched off and will be reinstalled from scratch (done the 2006-03-08):
    • lxn1177 (Prod RB)
    • lxn1186 (Prod RB)
    • lxn1188 (Test zone RB)
    • lxn1185 (CMS RB)
    • lxb2008 (EGEE.BDII for LHCB)
    • lxn1187 (EGEE.BDII for CMS)
    • lxn1189 (Test zone EGEE.BDII)

  • 2006-02-24: Need to restart two daemons on gdrb08. Fixed.

  • 2006-02-24: Raid disk full on gdrb08. Fixed.

  • 2006-02-24: gdrb01, gdrb02, gdrb03, gdrb09 and gdrb10 upgraded to LCG 2.7.0.

  • 2006-02-23: Beginning of the migration to LCG 2.7.0 on all the machines in production (gdrb11 updated).

  • 2006-02-22: monb001 (New monbox managed by FIO) tested. and is ok. Go in production.

  • 2006-02-20: VO Compass supported by gdrb01.

  • 2006-02-13: VO Compass supported by gdrb03.

  • 2006-02-09: LCG 2.7.0 installed on bdii103 and bdii104 (alias prod-bdii).

  • 2006-02-09: LCG 2.7.0 installed on bdii101 and bdii102 (alias lcg-bdii).

  • 2006-02-08: monb001 installed and configured as a new monbox (managed by FIO) with LCG 2.7.0.

  • 2006-02-06: raid disk full on gdrb01 and gdrb03. Fixed.

  • 2006-02-03: Problem with afs authentification on lxb1930. Need to restart ntpd service. Fixed.

  • 2006-02-02: gdrb04 is now using lcg-bdii as a BDII (instead of lxb2008).

  • 2006-02-02: Connection/routing problems with some sites in Taiwan and IN2P3. Fixed.

  • 2006-02-02: Alias lcg-bdii points now to bdii101 and bdii102.

  • 2006-01-27: Bug (known) in the myproxy-server daemon fixed on myproxy (the myproxy-server deamon was in a deadlock this morning).

  • 2006-01-26: RAID disk in degraded mode on gdrb03 (hdg dead ?).

  • 2006-01-26: Restart nfs service on lxb2003 (SE for LHCB).

  • 2006-01-25: Restart nfs service on lxb2004 (UI for LHCB).

  • 2006-01-25: Power supply on lxn1181 (myproxy.cern.ch) failure. Swith this service to node lxn1192.

  • 2006-01-25: Power cut in the CERN CC last night.

  • 2006-01-25: Need to restart services /edg-wl-lm, edg-wl-jc and edg-fmon-server on gdrb03.

  • 2006-01-24: lxn1192 (archiver) removed from production. Reinstalled from scratch and put it as free.

  • 2006-01-22: New CE available for CERN-PROD: ce101. All the RBs will point to it when LCG 2.7.0 will be installed.

  • 2006-01-20: /var partition almost full on gdrb04 due to a very huge file (/var/wtmp.1). Fixed by bzip2'ing it.

  • 2006-01-19: Hard disk changed on lxn1194. Reinstalled from scratch and put it as free.

  • 2006-01-19: Memory changed on gdrb10. This machine goes back in production.

  • 2006-01-13: Problem with the interlogd daemon on gdrb04. Fixed by David.

  • 2006-01-13: There was some trouble with the aliases lcg-bdii and prod-bdii but it has been fixed this morning. To sum up:
    • lcg-bdii points to: bdii001 and bdii002.
    • prod-bdii points to: bdii103 and bdii104.
    • Alias site-bdii should point to prod-bdii (not yet).

  • 2006-01-12: bdii103 and bdii104 are now prod-bdii machines (it corresponds to site-bdii).

  • 2006-01-10: lxn1178 and lxn1192 blocked. Need to reboot these two machines. Fixed.

  • 2006-01-06: Problem with the raid disk on gdrb03. Machine rebooted and raid disks checked. Fixed.

  • 2006-01-06: Serious problem with hda on lxn1194 (site-bdii.cern.ch). Machine removed from production and ITCM ticket generated.

  • 2006-01-06: gdrb10 freezes at random time. Machine removed from production and ITCM ticket generated.

  • 2006-01-06: Problem with the raid disk on gdrb03. Need to reboot the machine. No more error detected.

  • 2006-01-05: Emergency power cut of the computing center affecting all services. Services back at the end of the afternoon.

  • 2006-01-03: IO errors with the raid disk on gdrb06. Services stopped. Machine rebooted in the evening, and no more error detected.

  • 2006-01-03: R-GMA developers have now root access to all the production RBs (gdrbxx).


Some bugs

  • [[https://savannah.cern.ch/bugs/?21909][#20973]: WM crashes on multiple anyMatch requirements (LCG RBs).
  • [[https://savannah.cern.ch/bugs/?21909][#21909]: glite-wms-check-daemons.cron needs to redirect stderr to /dev/null (gLite WMS).


Edit | Attach | Watch | Print version | History: r638 < r637 < r636 < r635 < r634 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r638 - 2008-08-05 - SteveTraylen
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LCG All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback