SRE Data Durability

Friday 29 Nov 2019, 14:00 → 15:00 Europe/Zurich

31/S-023 (CERN)

Maria Arsuaga Rios (CERN)

Follow up of EOS SRE data durability.

- Understand the issue of the case checksum 0 with filesize !=0 found in restic.

- Repair plain layout for versioning files (skipped in this moment).

- Think how to automatize better the cases as all rep corrupted to help to accelerate the draining with these kind of leftovers. When dropping the replicas the entry in the namespace is removed (confirmed test from Cristi).

- Change cronjob timing from 20:00 to 15:00 for automatic reparation in failed drain filesystems.

- However keep the three objectives as the top priorities (last slide)

• fsck tests pps -> backup -> cms

14:00 → 14:30 SRE Data Durability Current State and Future Plans

Speaker: Maria Arsuaga Rios (CERN)

EOS_SRE_DataDurability_2019_11_29.pdf

EOS_SRE_DataDurability_2019_11_29.pptx

14:30 → 15:00 Discussion and Actions