HEPTape Web>Survey>CCIN2P3_Survey (2018-04-19, OliverKeeble)

Site

Site and Endpoints
Question	*________________________________Response__________________________________*
What is the site name?	CC IN2P3
Which endpoint URLs do your archival systems expose?	For Atlas / CMS / LHC (dCache) srm://ccsrm.in2p3.fr For Alice (XRootd) root://ccxrdralice.in2p3.fr:1096/
How is tape storage selected for a write (choice of endpoint, specification of a spacetoken, namespace prefix).	in dCache, we have different spacetokens used to select tapes pools. The XRootd endpoint for alice is tape only
Queue
What limits should clients respect?
---> Max number of outstanding requests in number of files or data volume	> 100 K
---> Max submission rate for recalls or queries
---> Min/Max bulk request size (srmBringOnline or equivalent) in files or data volume	Files: Max : 100 K Min : 1 K Volume > 100 TB
Should clients back off under certain circumstances?
---> How is this signalled to client?	SRM_INTERNAL_ERROR at request level and SRM_FILE_BUSY at file level returned by SRM. Stalling client by xrootd.
---> For which operations?	For SRM, any synchronous operation. For xrootd, any operation can be stalled by the server.
Is it advantageous to group requests by a particular criterion (e.g. tape family, date)?
---> What criterion?	Group requests by creation time in dcache: Data written in the same time are grouped on the same tapes. Reading data according creation time wil help to reduce mount/dimount of the sames tapes.
Prioritisation
Can you handle priority requests?	No, tape archive is shared between all VO and we not handle priority. But all recall request coming from dCache and Xrootd take benefit of our tape queuing system (TREQS : Tape Request Scheduler)
---> How is this requested?
Protocol support
Are there any unsupported or partially supported operations (e.g. pinning) ?
Timeouts
What timeouts do you recommend?	Timeout should be increase to 24h in order to benefit of large bulk recall
Do you have hardcoded or default timeouts?	Default dcache timeout per request on tape pool is 14400 s (4h)
Operations and metrics
Can you provide total sum of data stored by VO in the archive to 100TB accuracy?	Yes, accounting value is computed in byte
Can you provide space occupied on tapes by VO (includes deleted data, but not yet reclaimed space) to 100TB accuracy?	Yes, it is the same value as above.
How do you allocate free tape space to VOs?	We monitor the storage class usages of all VOs, and we do the allocation by bunch of 50-100 tapes when a storage class goes short of tapes
What is the frequency with which you run repack operations to reclaim space on tapes after data deletion?	We run repack manually when tape filling is bellow 70-80 %. We also run repack when we suspect tape to generate errors on recall
Recommendations for clients
Recommendation 1	Run prestaging (ie SRM BRINGONLINE) on large dataset with an huge timeout value.
---> Information required by users to follow advice
Recommendation 2
Buffer Management
Should a client stop submitting recalls if the available buffer space reaches a threshold?
---> How can a client determine the buffer used and free space?
---> What is the threshold (high water mark)?
---> When should the client restart submission (low water mark)?
If the client does not have to back off on a full buffer, and you support pinning, how is the buffer managed?
---> Is data moved from buffer to another local disk, either by the HSM or by an external agent?
Additional questions
Should any other questions appear in subsequent iterations of this survey?

-- OliverKeeble - 2018-01-30

Topic revision: r3 - 2018-04-19 - OliverKeeble

HEPTape

Public webs

- Cern Search
- TWiki Search
- Google Search
HEPTape All webs

Copyright &© 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback