OSG Technology Area Meeting, 10 August 2021¶
- Coordinates: Conference: +1-415-655-0002, PIN: 146 266 9392, https://morgridge.webex.com/webappng/sites/morgridge/meeting/info/791d9dddc5464eb6a73fc7746331d06c (password sent separately)
- Attending: BrianL, Carl, Marian, Mat, TimT
Announcements¶
- OSG User School in session until the end of the week
Triage Duty¶
- This week: Mat
- Next week: TimT
- 9 (+2) open FreshDesk tickets
- 0 (+0) open GGUS ticket
Jira (as of Monday)¶
# of tickets | Δ | State |
---|---|---|
155 | +2 | Open |
26 | +0 | Selected for Dev |
39 | +0 | In Progress |
6 | -5 | Dev Complete |
8 | +3 | Ready for Testing |
6 | +1 | Ready for Release |
OSG Software Team¶
- Kuberentes Hackathon today 1-5pm
- Tim makeup doc focus: EL8 interest is increasing
- Doc focus Aug 20
- Release
- AI (TimT): SciTokens update for EL8 (SOFTWARE-4126)
- AI (Carl): Improve default ProbeConfig directory configuration (SOFTWARE-4621)
- AI (BrianL): XRootD in OSG 3.6 (SOFTWARE-4494)
- GSI Transition
- AI (Carl): osg-token-renewer package
- AI (Mat): Add API token authentication to Topology (SOFTWARE-4134)
- AI (BrianL): Get XRootD tests running with proxies again (SOFTWARE-4604)
- Other
- AI (BrianL): Investigate persistent condor_ce_run failures in 3.6 nightlies
- AI (Mat): Review comments on cvmfsexec doc changes
- AI (BrianL): Add CI tests
Discussion¶
- GRACC upgrade is planned for the end of August
Support Update¶
- Georgia Tech (BrianL, Mats): Need to verify that the config worked to prevent pilots from running in pilots
- Lancium (BrianL, Mats): their GPUs are still going unused, it seems that
RecentJobStarts
is frequently at its limit of 400 starts/20 min - JLab (Marian): Fixed HTCondor configuration; JLab admins should be informed to expect some downtime and manual changes between OSG release series upgrades
- Lamar (Marian): Jobs are running but requesting Software Team help to examine job classads for anything unexpected
- BNL (Marian): Troubleshooting with John de Stefano and other BNL admins; getting unexpected behavior from Freshdesk with notes or replies not getting to everyone on the ticket. Marian will notify Freshdesk support about the strange behavior.
- Hosted CEs (Carl): Investigate hosted-ce35 not reporting to GRACC.
- University of Illinois (Mat): Undesired XRootD HTTP plugin behavior; Mat is following discussion between XRootD developers
OSG DevOps¶
New: - Request to inject entire HTCondor job classad into GRACC records. Will require gratia-probe-condor changes.
Ongoing: - Elasticsearch will need to be updated, scheduled for last week in August. - StashCP go client is working and tested, and being tested by Christina now. Need to talk to Mats on packaging and releasing. (will follow up). No news from Christina, - Design document is done and is being distributed for xrootd monitoring collector NG (next gen). Shoveler is working and lightly tested. Collector is 80% done. Design document is still not reviewed, but moving foward anyways. - Discussions of xrootd-multiuser support for checksumming on-the-fly. After a discussion, the checksum-on-the-fly funcationality will be ported to xrootd-multiuser. Copied largely from xrootd-hdfs. This does mean that out-of-order writes will not be supported.
Back burner: - (Stalled, but need to get back to it) XRootD accounting information (from Frank): - May need help with automating the process of generating these reports. A month only takes a few minutes. Longer periods take much longer. - Table of working set, read, and re-read for monthly, quarterly, and year. - XRootD TCP plugin is ready for packaging. Derek will follow up with software team on who to hand that off to. - https://opensciencegrid.atlassian.net/browse/SOFTWARE-4744
Discussion¶
None this week
OSG Release Team¶
- OSG 3.5.45 and OSG 3.6 (Thursday)
- Ready for Release
- gratia-probe 1.24.0 (OSG 3.5 and 3.6)
- XRootD 5.3.1 (OSG 3.5-upcoming)
- Ready for Release
Discussion¶
None this week