US CMS Tier-2 Facilities Deployment Status

This page is used to track the status of upgrading various products and services at the US CMS Tier-2 sites. Please update your status when you complete an upgrade.

2019 Upgrades and Milestones:

  • Complete any lagging 2018 milestones and upgrades (proposed completion date: June 30, 2019):
    • UCSD:
      • Upgrade all WN's to SL7
    • MIT:
      • Upgrade all WN's to SL7
      • Storage dual-stack
      • Xrootd IPv6
  • Upgrade WN's to RHEL 7.6+ (December 31, 2019)
  • Unprivildged Singularity deployed at one site (September 30, 2019)
  • Upgrade to HTCondor 8.8.2+ (September 30, 2019) - SUSPENDED UNTIL AFTER OSG 3.5 RELEASE
  • Upgrade to OSG 3.4.28+ (September 30, 2019), which includes Xrootd 4.9.1, and at least frontier-squid 3.5.28-3 and cvmfs 2.4.3.
  • Read/write/stageout via Xrootd software; HTTP-TPC transfers work (December 31, 2019). Ensure the XRootD services at the site support both HTTPS and xrootd protocols for reading and writing. Ensure stageout occurs to XRootD service via HTTPS protocol (not GridFTP server). Ensure that HTTP-TPC works with the XRootD services; perform PhEDEx transfers to Nebraska via HTTP-TPC and participate in the WLCG DOMA TPC-based tests. See https://twiki.cern.ch/twiki/bin/view/Main/XRootDoverHTTP

Please input the completion date in the table below:

Upgrade Caltech Florida MIT Nebraska Purdue UCSD Wisconsin Vanderbilt SPRACE
Unprivildged Singularity 05/09/2019     07/23/2019 05/15/2019*        
Upgrade WN's to RHEL 7.6+. 05/01/2019 partial (1/3) partial (50%) done 01/22/2019   4/1/2019   done
Upgrade to OSG 3.4.28+ 05/28/2019   partial (40%) 07/23/2019 7/23/2019   5/16/2019   done
Upgrade to frontier-squid 3.5.28-3+ 05/28/2019 07/26/2019 done done 7/25/2019   5/1/2019    
Upgrade to cvmfs 2.4.3+ 04/11/2019 07/24/2019 done 01/22/2019 7/23/2019   5/16/2019   done
Upgrade to HTCondor 8.8.2+ 09/04/2019     done          
Read/write/stageout via Xrootd software over HTTPS protocol 05/09/2019 01/01/2019         5/16/2019    
HTTP TPC transfers work 05/03/2019 04/25/2019              
Carry-overs from 2018 N/A N/A   N/A N/A   N/A N/A N/A
Whole-node scheduling (%) 12/21/2018 No plan   done 60%        
* CMS dedicated resources run in unprivileged mode, opportunistic slots do not

? Does 'http transfers work' mean xrootd 4.9.1 plus configs ? (bockjoo)

Goals (not milestones)

  • Whole-node scheduling (i.e. maximize cores in pilots)
  • Participation in WLCG DOMA project (davs protocol)

Performance Metrics:

  • Maintain at least 90% Site Availability and Site Readiness for the calendar year (December 31, 2019)

Longer-term Planning

  • Try out RHEL8 (depending on upstream OSG software release)
  • OSG 3.5 (depending on upstream OSG software release)
  • From IRIS-HEP: With OSG, help transition 30% of data transfers at one USLHC site to use a non-Globus Toolkit implementation (31 March 2020).
  • Retire PhEDEx (2020)
  • Retirement of gFTP (medium-term - no roadmap yet)
  • Investigate OSG and IRIS-HEP K8s service deployments (medium-term). Hosted CE’s?
  • Investigate OSG xrootd federation (medium-term)
  • Moving away from x509 authentication (long-term)

Why We Upgrade

No site exists in a vacuum. We exist within the wider ecosystems of the WLCG, global CMS, and our national organizations such as U.S. CMS. Some of the upgrades we are asking you to do are on the critical path for other upgrades, deployments, or service retirements, such as enabling dual-stack or stageout via https.

Other upgrades are tied to the retirement of dependent software packages or operating systems. Scientific Linusx 6 and CentOS 6 have an end-of-life in 2020, approximately two years from now. Moving services and worker nodes is important to do now, well ahead of the end of support for such things as security patches, so there are no surprises. A blocker that is found at a site may take a while to resolve. The risk of an interruption of service motivates us to stay ahead of the curve sufficiently to maintain excellent service. OSG 3.3 end-of-life was already in May 2018. Support for PerfSONAR 4.0 and all support for PerfSONAR on SL6 will end in early 2019.

Finally there are upgrades that we ask you to do to make site architecture simpler and more secure, such as retiring SRM, GUMS, pool accounts, and the OSG WN client. Maintaining extraneous software packages takes operational effort, and we in U.S. CMS want to conserve effort as much as possible by making sites easier to maintain.

We have not prioritized upgrades beyond providing guidance on completion dates in the table below. However, we do want all of the upgrades completed by the end of calendar 2018. In particular, the last one or two sites to complete an upgrade may be blocking a downstream upgrade or research project.

2018 Upgrades and Milestones:

Please would sites update the following table when upgrades are done by entering the date of completion. Estimates of when you plan on accomplishing a particular upgrade are also welcome, but please precede it with "Est." Older carryover milestones from before January 2018 can be found lower down on the table. Details of the upgrades can be found below the table. Upgrades in orange have been postponed due to lack of software releases upstream. One upgrade is optional (retire OSG WN Client). Worker nodes may take some time to upgrade to SL7 since storage systems are quite full and many of our sites distribute storage on the worker nodes. Without buffer space it is difficult to upgrade since nodes must be cleared. Note that there are still 2017 upgrades to be completed by some sites!

Upgrade Complete By Caltech Florida MIT Nebraska Purdue UCSD Wisconsin Vanderbilt SPRACE
Singularity: OS=any 04/01/18 04/03/18 04/01/18 done done <04/01/18 <04/01/18 04/11/18   <04/01/18
Retire glexec 04/01/18 <04/01/18 04/01/18 done done <04/01/18 <04/01/18 <04/01/18   <04/01/18
OSG 3.4 across site May <04/01/18 04/01/18 done done 11/14/18 done done c7 update in progress done
HDFS 3.x (if OSG releases) not released   NA           NA NA
Upgrade all CEs to SL7 start of run done done done done 06/04/18 done done done for C7 cluster done
Upgrade all squids to SL7 start of run 05/13/18 done done done 08/20/18 done 06/20/18    
Upgrade all xrootd to SL7 start of run 05/13/18 done done done 07/07/18 done done done partially done
Upgrade all gFTP to SL7 start of run 05/13/18 done done done 05/25/18 done done done partially done
xrootd 4.8.2 05/01/18 29/06/2018 done (4.8.3) done done done (4.8.5) 11/14/18 done done (4.8.5) done done
Retire GUMS (lcmaps) 06/01/18 done done done done 07/12/18 done 09/17/18 done done
Retire pool accounts 06/01/18 done done done done 11/14/18 done 09/17/18 done done
perfSONAR 4.1 on SL7 10/01/18 done done done done 10/11/18 done 09/14/18   done
Upgrade WNs to SL7 12/31/18 19/11/18 done in progr done 06/04/18 25% done done in progress done
Retire OSG WN client Optional 19/11/18 done done done       done  
Stageout via HTTPS 12/31/18 done done done done 12/04/18 done done need plugin  
HTCondor 8.8.x not released   NA           NA  
                     
Storage dual-stack 12/31/17 done 22/02/19 done   done done done done done done
Upgrade one CE to SL7 12/31/17 29/06/2018 done done done done done done done done done
xrootd HTTPS 12/31/17 done done done done done done done   done
xrootd IPv6 12/31/17 done 22/02/19 done   done done done done done done
xrootd >4.6.x 12/31/17 done done done done done done done done done
cvmfs 2.4 12/31/17 done done done done done done 04/01/18 done on C7 done
HTCondor 8.6.x 12/31/17 done NA done done done done done NA done
Load-balanced gFTP 12/31/17 05/13/18 done done done done done done done done NA
Notes:
  • No information on when MIT campus infrastructure will support IPv6
  • HTCondor 8.8.0 will be released only in November 2018. This upgrade will be postponed until 2019.
  • HDFS 3.x has not yet been released by OSG. No ETA.

Progress US CMS HEP - 4Q 2018: Completed cvmfs 2.4 ('17), xrootd HTTPS (also '17), and Upgrade all squids to SL7

Details of Upgrades:

  • Deploy 2018 Tier-2 resource pledge by April 1, i.e. provide 25% of 2018 CMS Tier-2 resource request as needed to meet data-taking demands. This means each site should have installed 2,500 TB of storage by April 1, 2018 and 2,800 TB by April 1, 2019 (but effectively by the end of calendar 2018). Current deployment at each site meets the processing resource pledge.

  • Performance goals:
    • Maintain site availability and readiness above 80% in accordance with CMS requirements - December 31st
    • Our US CMS goal should be 90%

  • Storage (GridFTP) and xrootd are IPv6 accessible. See deployment campaign twiki.
  • Once Singularity is put into production in the CMS Global Pool and the SAM tests are modified:
    • Retire glexec
    • Upgrade worker nodes to SL7
    • OSG worker node client might not be needed anymore, since everything ships with the container, except that some SAM tests still run only on the bare machine, not within the containers. We need containered versions of these tests - TBD, before end of 2018
    • Retire pool accounts and GUMS (June 1st): Note that GUMS support ends June 1st. The fallback solution is to verify that GUMS isn’t depending on VOMS-Admin. See also documents on VOMS Admin server retirement github and transition to LCMAPS VOMS github.
    • Set GLIDEIN_REQUIRE_OS=any
  • Upgrade HDFS: Still waiting on upstream for HDFS 3.0. Error correcting codes should increase usable space at sites by ~30% “for free.” To avoid upgrading twice within a year, stay on current version for 2017. [Q: Tied to OSG 3.4 upgrade?] - Date TBD
  • Track quality of all data reported to WLCG - currently only tracking wall clock time. [Investigations in January 2018 found lots of missing (zeroed) CPU time data from Nebraska and Purdue, as well as other sites world-wide, which affects the CMS CPU efficiency metric.] - February - DONE Feb 6
  • Migrate perfSONAR to version 4.1 on SL7. OSG documentation - October 1. Current version (10/25/18) is 4.1.2-1.el7.

In addition, there are some incomplete upgrades carried over from 2017, which we would like done by April 1st:

  • Migrate to cvmfs 2.4 (only Caltech has done it)
  • One site (Caltech) still needs to implement load-balanced gridFTP
  • One site (UCSD) still needs to upgrade to XrootD 4.6.x
  • Two sites still need to upgrade at least one CE to SL7
  • Three sites still need to upgrade to HTCondor 8.6.x
  • Three sites still need to upgrade to HTTPS XrootD
  • Five sites still need to upgrade to version 3.4 of the OSG software stack.

Random ideas:

  • Move to whole-node scheduling (for security) ?

2017 Upgrades and Milestones:

Performance goals for the entire year:

  • Maintain Site Availability above 80% in accordance with WLCG requirements.
  • Maintain Site Readiness above 80% in accordance with CMS requirements.
  • Should these be increased to 90% goals?

General milestones (for the L2's, not the sites):

  • Establish a metric of perfSonar availability. (by July)
  • Verify for each site the consistency of the information in the Resource Deployment Table, the HS06 spreadsheet, CPU models and DB12 fast benchmarks obtained from the Global Pool, and OIM. (by December)

Site-specific upgrades and milestones:

  • Deploy 2017 Tier-2 resource pledge (by April 1; done in January). Provide 25% of increased CMS Tier-2 resource request as needed to meet data-taking demands
  • Install Singularity (by April 1; done in March)
  • Deploy load-balanced gftp and retire bestman2 (by September; 3/10 sites completed as of March 2017)
  • Install new spacemon client for regular storage dumps (once/week). Note that there is a new set of dump scripts for hadoop filesystems. (by May; 4/7 sites completed by March: Florida, Nebraska, Purdue, Wisconsin)
  • Upgrade to HTCondor 8.6.x (by July)
  • Upgrade to OSG 3.4 (estimated release date: June 2017, complete by December 2017).
  • Update of /store/unmergered, /store/temp, and /store/user/temp cleaning policy to either 4 weeks from 2 weeks, or
implement exception list. (by May, already done?)
  • Change site-local-config.xml to use gfal2 not deprecated lcg-utils for stage out. (by July)
  • Upgrade to cvmfs 2.4: Will bring “external caches” — keep CVMFS caches on local disk and/or HDFS (Lustre). (by October)
  • Upgrade to XRootD 4.6.x: Yields important performance gains to proxy cache component. (by October)
  • Deploy MJF file with HS06 benchmark information (by October)

Carried over upgrades and milestones from previous years:

  • Retire GIP/BDII (hard deadline is April 1; already done?)
  • AAA or xrootd-related milestones:
    • Deploy IPv6 compliant xrootd - carried over from 2015 (4/10 sites completed by March 2017: Caltech, Nebraska, Purdue, Wisconsin)
    • Export HTTPS protocol via xrootd software package - towards building http-access for AAA. (3/10 sites completed by March 2017: Caltech, Nebraska, Purdue)
  • Convert at least one CE at each site to RHEL7 (4/10 sites completed by March 2017: Nebraska, Purdue, UCSD, Wisconsin)

Milestones for 2018 (or later in 2017):

  • Once Singularity is put into production in the CMS Global Pool and the SAM tests are modified:
    • Retire glexec
    • Upgrade workernodes to SL7
    • OSG worker node client might not be needed anymore, since everything ships with the container (TBD, 2018?)
    • Retire pool accounts and GUMS? Discussion with CMS underway.
  • Upgrade HDFS: Still waiting on upstream for HDFS 3.0. Error correcting codes should increase usable space at sites by ~30% “for free.” To avoid upgrading twice within a year, stay on current version for 2017.

Site BDII Sing. load-bal. gftp spacemon HTCondor 8.6.x OSG 3.4 /store /unmerged gfal2 cvmfs 2.4 XRootD >4.6.x IPv6 xrootd HTTPS xrootd CE to SL7
T2_BR_UERJ                          
T2_BR_SPRACE Yes / Done Yes / Done   Yes / Done       Yes / Done   Yes / Done Yes / Done Yes / Done  
T2_US_Caltech Yes / Done Yes / Done New HW is here Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done NW upgrade in Dec Yes / Done In test
T2_US_Florida Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done   Yes / Done IPv6 on Nov's end Yes / Done Yes / Done
T2_US_MIT Yes / Done Yes / Done Yes / Done Yes / Done     Yes / Done Yes / Done   Yes / Done      
T2_US_Nebraska Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done
T2_US_Purdue Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done   Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done
T2_US_UCSD Yes / Done Yes / Done Yes / Done Yes / Done     Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done
T2_US_Wisconsin Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done   Yes / Done Yes / Done   Yes / Done Yes / Done   Yes / Done
T2_US_Vanderbilt Yes / Done Yes / Done Yes / Done Yes / Done     Yes / Done Yes / Done         Yes / Done

2016 Upgrades and Milestones

General milestones (not driven by the sites):

  • Establish a metric of perfSonar availability
  • Complete remaining connections to LHCONE
  • Please keep the Resource Deployment Table up-to-date as you purchase or retire hardware.

Site-specific upgrades and milestones:

  • Decommission legacy components - carried over from 2015 :
    • BDII - partially done at Nebraska.
    • GRAM CEs
  • Upgrade to OSG 3.3 - before August
  • For at least 3 sites, implement load-balanced GridFTP servers via DNS or IP - prelude to decommissioning bestman2
  • Improve the consistency of HS06 benchmarking and APEL normalization constants.
  • Commision multicore pilots at all US Tier-2s (April 1) - on track to complete
  • Upgrade to HTCondor 8.4.x - carried over from 2015 (3/10 sites completed).
  • Convert at least one CE at each site to RHEL7
  • AAA or xrootd-related milestones:
    • Deploy IPv6 compliant xrootd - carried over from 2015 (3/10 sites completed)
    • Export HTTPS protocol via xrootd software package - towards building http-access for AAA.
    • Upgrade WNs to cvmfs 2.2.2 or later - allows us to export AAA data federations

Site BDII GRAM OSG 3.3 Load-balanced gridFTP HS06/APEL Multi-core HTCondor 8.4.x CE to RHEL7 IPv6 xrootd HTTPS xrootd cvmfs 2.2.2+
T2_BR_SPRACE No Yes / Done Yes / Done No No No Yes / Done No No No Yes / Done
T2_BR_UERJ No No No No No No No No No No No
T2_US_Caltech Yes / Done Yes / Done Yes / Done No Yes / Done Yes / Done Yes / Done No Yes / Done Yes / Done Yes / Done
T2_US_Florida No Yes / Done Yes / Done No Yes / Done Yes / Done Yes / Done No No No No
T2_US_MIT No Yes / Done Yes / Done No No Yes / Done Yes / Done No No No Yes / Done
T2_US_Nebraska Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done
T2_US_Purdue Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done
T2_US_UCSD Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done No No Yes / Done
T2_US_Vanderbilt No Yes / Done No No No Yes / Done No No No No No
T2_US_Wisconsin Yes / Done Yes / Done Yes / Done No Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done No Yes / Done

2015 Table

Site Benchmark HS06 Shutdown GRAM CEs Shutdown BDII IPv6 xrootd HTCondor 8.4 Upgrade Improve perfSonar performance LHCONE Deployment Multi-core Last Update
T2_BR_SPRACE No Yes / Done No No Yes / Done No   No 2016-02-01
T2_BR_UERJ No No No No No No   No 2015-08-21
T2_US_Caltech Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done Yes / Done DONE Yes / Done  
T2_US_Florida DONE DONE
(Monitor only)
No No No No DONE No 2015-10-24
T2_US_MIT No DONE No No DONE No DONE No  
T2_US_Nebraska No DONE Partial (testing) DONE No No DONE No  
T2_US_Purdue DONE DONE No DONE No No DONE DONE 2015-10-02
T2_US_UCSD No Yes / Done No No Yes / Done No DONE No 2016-09-20
T2_US_Wisconsin DONE DONE No DONE DONE No DONE No 2016-03-11
T2_US_Vanderbilt No Yes / Done No No No No No No  
Notes on each of the goals:
  • Benchmark HS06: Buy a software license and benchmark all machine types at your site, and update in the CPU tallying spreadsheet.
  • Shutdown GRAM CEs for CMS: These were replaced by HTCondor CEs. We want all CMS pilots to enter sites via HTCondor CEs only, not GRAM CEs.
  • Shutdown BDII: (need more details) <- what is the reason?, just curious
  • IPv6 xrootd: deploy IPv6 compliant xrootd (need more details, recommended version)
  • HTCondor 8.4 Upgrade: Estimated to be available Summer 2015.
  • Improve perfSonar performance: Performance can be viewed here.
  • LHCONE Deployment: Connect to ESNet's LHCONE VPN. Your site will be contacted by ESNet.
  • Multi-core scheduling: While not obligatory this year, eventually we will need this to run multi-threaded workflows. If you have plans to move to multi-core this year, please let us know.

Other goals:

  • Accounting Review:
  • Implement Local Access:

2014 Table

Site Finish 2013 table OSG 3.2 Upgrade HTCondor-CE Config Management WN IPv6 Services IPv6 >20Gbps test AAA 2k clients AAA 4k clients HTCondor 8.2 Last Update
T2_BR_SPRACE DONE DONE Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown July, 29 2014
T2_BR_UERJ DONE DONE DONE DONE via kickstart Not planned Not planned No No No Unknown Never
T2_US_Caltech DONE DONE DONE DONE as needed Q1 DONE DONE DONE DONE 2014-12-16
T2_US_Florida DONE DONE DONE DONE via image Unknown Unknown DONE DONE DONE N/A 2014-11-17
T2_US_MIT DONE DONE DONE DONE if needed if needed No DONE DONE DONE 2015-03-06
T2_US_Nebraska DONE DONE DONE DONE DONE Partial DONE DONE DONE DONE 2014-08-26
T2_US_Purdue DONE DONE DONE DONE Partial Partial DONE DONE DONE DONE 2014-2-17
T2_US_UCSD DONE DONE DONE DONE Ready Ready Q3 DONE DONE DONE 2015-02-06
T2_US_Wisconsin DONE DONE DONE DONE DONE Ready DONE DONE DONE DONE 2015-02-16
T2_US_Vanderbilt DONE DONE DONE DONE via image Never If truly needed No DONE DONE N/A 2014-1-5
Notes on each of the goals:
  • Finish 2013 table: All entries in the 2013 table are complete.
  • OSG 3.2 upgrade: All service nodes (GUMS, CE, SE) and worker nodes are running a OSG 3.2 release.
  • HTCondor-CE: One CE at the site has the glideinWMS submitting pilots through the HTCondor-CE. Turning off GRAM (which cannot be done until after gliteWMS is disabled) will be made into a deadline later.
  • Config management: The CEs at your site should be configured using a piece of configuration management software (puppet or chef, for example) and not through homegrown scripts. Before checking this off, the site should conduct a "fire drill" where the CE disk image is reformatted and allow the configuration management to restore it without admin intervention.
  • WN IPv6: All worker nodes on the T2 cluster should have outgoing IPv6 connectivity.
  • Services IPv6: The public-facing services (in particular: Xrootd, SRM, and GridFTP) should listen to IPv6 sockets and conduct transfers via IPv6.
  • >20Gbps test: Demonstrate the ability to do disk-to-disk transfers at your site at a rate greater than 20Gbps through the WAN using the PhEDEx Load Test instance.
  • AAA 2k clients: Have the AAA team demonstrate the ability to sustain 2,000 simultaneous client applications against your site's Xrootd install.
  • AAA 4k clients: Have the AAA team demonstrate the ability to sustain 4,000 simultaneous client applications against your site's Xrootd install.
  • HTCondor 8.2: Upgrade your site's HTCondor batch system to 8.2.0 or later. The 8.2.0 release is expected in May 2014.

2013 Table

Site OSG3 Aux HDFS 2.0 DigiCert SL6 Enable t1production role Adjust VOMS priorities Upgrade to perfSONAR 3.3.1 Upgrade to perfSonar 3.3.2 Enable OASIS SHA-2 compliant OSG CVMFS 2.1.15 perfSONAR WLCG and USCMS meshes Last Update
T2_BR_SPRACE DONE N/A N/A DONE DONE DONE DONE DONE DONE DONE DONE DONE July 29, 2014
T2_BR_UERJ DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE End of March Feb 11, 2014
T2_US_Caltech DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE April 3, 2014
T2_US_Florida DONE N/A DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE Feb 14, 2014
T2_US_MIT DONE End Feb DONE DONE DONE DONE DONE DONE DONE DONE   DONE 11-Feb-2014
T2_US_Nebraska DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE Feb 11, 2014
T2_US_Purdue DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE May 21, 2014
T2_US_UCSD DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE March 10, 2014
T2_US_Wisconsin DONE DONE DONE DONE DONE DONE DONE DONE DONE DONE   DONE Mar 10, 2014
T2_US_Vanderbilt DONE N/A DONE DONE N/A N/A DONE DONE DONE DONE DONE DONE 15 Apr 2014
Notes on the states:
  • 11-Feb-2014 mtiernan@lnsNOSPAMPLEASE.mit.edu - Added column for upgrade to perfSonar 3.3.2
  • "Unknown" means the status was not known when this table was bootstrapped by Brian.
  • "Partial" means some of the upgrades have already happened.
  • If an item has a No, then please put in the estimated upgrade date.
Notes on the upgrades:
  • OSG3 CE: all CEs are on OSG3.
  • OSG3 Worker nodes: all WNs are on OSG3
  • OSG3 Aux: All auxiliary services (xrootd, gridftp, SRM) are based on OSG3.
  • HDFS: Upgrade to Hadoop 2.0. Please coordinate this item ahead of time with Brian.
  • HS06 spreadsheet: Update Ken Bloom's HS06 Google Docs spreadsheet
  • CVMFS: Switch CMS software distribution to CVMFS.
  • SL6: Upgrade worker nodes to SL6. Note that all grid nodes may also but upgraded to SL6, but we are not yet tracking this.
  • GUMS: Upgrade GUMS to be OSG3 RPM-based.
  • DigiCert: Switch site grid certificates to DigiCert. Please contact OSG Security if you need significantly more than 20 certificates.
  • Register perfSonar: Register your perfSonar instance with the GOC.
  • Enable t1production role: As stated; it should get mapped onto the account that the regular production role gets mapped to.
  • Enable file monitoring: See GenericFileMonitoring.
  • Adjust VOMS priorities: Set to 50% production/t1 production, 40% pilot, 10% all other roles.
  • Update squid: Implement latest version with proper monitoring enabled.
  • Upgrade to perfSonar 3.3: Available from http://psps.perfsonar.net/.
  • Enable OASIS. Install the oasis-config RPM from OSG and verify you can list /cvmfs/oasis.opensciencegrid.org from your worker nodes.
  • Last Update: Set this to the last date someone from your site reviewed/updated this table.

Responsible: JamesLetts

Edit | Attach | Watch | Print version | History: r385 < r384 < r383 < r382 < r381 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r385 - 2019-09-16 - GarhanAttebury
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic 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