Strange slowdown during ATLAS migration. Sebastien Masson reports: "for the query on the 17th around 12:00 there was no parallel used at all. Actually the database downgraded the Degree Of Parallelization to 1. The reason I can find is “DOP downgrade due to insufficient number of processes”." Awaiting result of further investigations.
Recalls started on Tuesday evening.
CTA is taking part with 4 PB, no multi-hop.
Some errors from the MGM: mostly these are rate-limiter warnings. A handful of errors to be investigated.
Julien will set up more CI runners but needs to configure them.
New CI setup will be to run only the basic tests by default, but to have all tests available to run manually for those who want them. Anyone who needs something more complicated can use the gitlab API to customise it further.
All tests will run nightly on the master branch.
Test concurrent archival, retrieval, deletion ("the mutex test")