Ceph

I'm not sure how that's possible. All objects in an erasure coded pool (3.59s0 shard shows it is one because of the trailing s0) are supposed to have hinfo_key attributes. Could this come from a newly added tests that is over zealous ?

description: upgrade:firefly-x/stress-split-erasure-code/{0-cluster/start.yaml 1-firefly-install/firefly.yaml
  2-partial-upgrade/firsthalf.yaml 3-thrash/default.yaml 4-mon/mona.yaml 5-workload/ec-rados-default.yaml
  6-next-mon/monb.yaml 8-next-mon/monc.yaml 9-workload/ec-rados-plugin=jerasure-k=3-m=1.yaml
  distros/ubuntu_12.04.yaml}

Loic Dachary wrote:

I'm not sure how that's possible. All objects in an erasure coded pool (3.59s0 shard shows it is one because of the trailing s0) are supposed to have hinfo_key attributes. Could this come from a newly added tests that is over zealous ?

This suite is new in part that it runs all workloads in parallel with upgrade.

Ok. Then I guess the problem is to find which workload is walking all objects and complaining about "hinfo_key". In http://qa-proxy.ceph.com/teuthology/teuthology-2014-12-10_17:13:01-upgrade:firefly-x-next-distro-basic-multi/647031/teuthology.log there only is one occurrence of the "hinfo" keyword. That's weird.

Any idea ?

Actually, the problem is likely that the hinfo_key was missing on whatever osd ended up as the "authoritative" osd. So, it's not an extra attr, it's a missing attr.

Same issue http://qa-proxy.ceph.com/teuthology/teuthology-2014-12-15_18:18:01-upgrade:firefly-x-giant-distro-basic-vps/661895/teuthology.log

Same issue two jobs failed 667642, 667643

Run http://pulpito.ceph.com/teuthology-2014-12-18_18:18:01-upgrade:firefly-x-giant-distro-basic-vps/

Same issue in run http://pulpito.ceph.com/teuthology-2014-12-21_17:13:02-upgrade:firefly-x-next-distro-basic-multi/

Job ['671674']