CTA Dev Meeting

Thursday 16 Mar 2023, 14:00 → 15:30 Europe/Zurich

513/R-068 (CERN)

Michael Davis (CERN)

14:00 → 14:10 CTA Release Roadmap

See CTA Release Roadmap

Release 4.8.6-0.rc1

• Release date: 24 Feb
• Pre-prod deployment date: TBD
• Prod deployment date: NOT
• Test encryption key ID changes performed by Vova
• Issue link

Release 4.8.6-1

• Release date: 8 Mar
• Pre-prod deployment date: 9 Mar
• Prod deployment date: 9 Mar
• Small release from branch "release_4.8.6"
• Fixed queue cleanup bug in CTA
• Milestone link

Release 4.8.7-1

• Release date: TBD
• Pre-prod deployment date: TBD
• Prod deployment date: TBD
• Waiting to include fix #330
• Milestone link

Public Release

• Latest version available on public repo: v4.7.14-1, v5.7.14-1
• For release v4.8.7-1/v5.8.7-1, wait one week to see if the production deployment has been stable.

Release 4.8.6-0.rc1

• Issue #330 is being tested here.
• We need to find out how the timeout value went from the config value to one second.
• As mentioned before, this is needed before release 4.8.7-1.

14:10 → 14:20 CTA dev topics

Improve "error_text" message in query prepare

• Discuss solution presented in #331

Add system tests for admin commands

• For more details #332.

FST checksum metadata error leads to whole archive batch failing

• For more details #ops-1022.

cta-taped deceiving read performance

• For more details #ops-1010.

"Needs discussion" topics

• Issues link

"Dev issue needed" topics

• Ops issues link

Improve "error_text" message in query prepare

• The &mgm.errmsg= parameter is specific to EOS (dCache has no concept of mgm). Therefore, we have to change this parameter if we want to use it.
• We need to confirm if the message body sent back to EOS contains any field with the error information. If so, we may not need to use the &mgm.errmsg= parameter.
• The &mgm.errmsg= part of the issue can be handled in a separate ticket. There is no need to mix it with the formatting of the "error_text" message.

Add system tests for admin commands

• Will be handled by Pablo.

FST checksum metadata error leads to whole archive batch failing

• Jaroslav will take care of it (also to get familiar with EOS part of CTA system). A new dev ticket will be created.
• We need to understand what is causing the race condition.

cta-taped deceiving read performance

• The problem - 2 second delay between file reads - is only visible with small files (for large files it becomes irrelevant).
• Vlado will check with experts about what is the underlying scsi problem.

Large number of CRITICAL errors in cta-taped - "In BackendPopulator::~BackendPopulator(): error deleting agent (cta::exception::Exception). Backtrace follows."

• Issue #1020.
• Lasse will try to continue debugging with, and also use it to learn better how the object store works.

14:20 → 14:30 Scheduler DB

Handover from David to Jaroslav

14:30 → 14:40 dCache Integration

AOBs

14:40 → 14:50 CTA dev board review

Objective

• Look at the active issues in our CTA dev board.
• Decide if they should be kept, removed, reassigned, prioritised, etc.

Review unlabelled issues

• Unlabelled issues list: link

Review "In progress" issues

• Full CTA board: link

Review specific topic

• Every week we review the issues of one topic.
• This week: "Scheduler" and "Object Store" labels

14:50 → 15:00 AOB

Other

• Room 513/R-068 is booked every week, until the EOY, for the CTA dev meeting.