Last modified 13 years ago Last modified on 10/08/07 10:58:34

dCache, SRM 2.2 deployment meeting Friday Oct 5, 2007

Participants : Flavia, Gene, Timur, Dmitry, Patrick

Thanks to Gene for the minutes :

  • First non-atlas sites will upgrade (fzk, in2p3), Then will ask atlas not to use one site at a time (while upgrading - rolling upgrade) because atlas is running a challenge.
  • Some sites have old releases, so removal of file won't reflect space goes back to token. Users should not see anything major, e.g. precious files will still go to tape after the upgrade.
  • Synchronizing switching to new paths may be a issue. This will be a topic for the GSSD meeting.

Flavia's Issues: Numbers refer to Flavia's implementation issues page at of today.

  • 2. Easy to fix - Timur
  • 5. 7, Dmitry need to double check this
  • 6. no road to cell - system was being restarted - Flavia will remove
  • 4. put status changes with time. Thinks we have fixed. looking for rply with flavia.
  • 10. BNKL not configured correctly. Works now. will mark fixed
  • 12. Flavia will remove (fixed)
  • 13. fixed but not verified. Here Dmitry meant to say that is should be fixed already and he is asking Flavia to rerun the tests against the fapl110 endpoint one more time. FIXED after test with DESY endpoint Mon Oct 8.
  • 14. fixed. not verified. This can't be tested on FAPL because the cms vo space is not configured. FIXED after test with DESY endpoint Mon Oct 8.
  • 19. Not critical, will fix later release
  • 28. Timur will look at this. Only if access latency is not specified. Flavia : not fixed with release 19 either : The client asks for REPLICA and it gets CUSTODIAL-NEARLINE.
  • 31. SRM prepare to put. (NOT FIXED according to flavia) According to Timur, the way the code is structured at the moment, it seems not to be a straightforward fix.
  • 32. SRM prepare to put.returns SRM_NO_FREE_SPACE on both request and file levels.

Quote Flavia

There is a new problem with srmAbortRequest that started to appear after the last changes and that is not reported in the twiki page. Dmitry knows about it. If I see it appearing again on Monday I will list it in the twiki page again.

According to Dmitry this has been fixed and committed. He will deploy the code on fapl shortly.

and the updated problem page

Summarize -

  1. GSSD discussion of synchronizing upgrade amongst Tier ones (path usage)
  2. All non-space reservations issues expected to be complete the end of Friday - Timur and Dmitry Space expected too.
  3. Cut and deploy at DESY Monday, Flavia will run tests and update the page so Fermi can present the residue at the GSSD.
  4. GIP discussion, we need to read glueschema v1.3 and provide comments. S. Burke will provide information on which fields in the Glue schema are used for the 'high' level tools and which are only interpreted by humans.
  5. All EU-Tier 1 will send an representitive to the Nov Workshop in Edinburgh. OSG will not attend the T1-T2 workshop

Last modified by patrick @ Sat Feb 27 01:58:35 2021