Giant Backports RGW failure

[Abhishek L <abhishek.lekshmanan@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1323 bytes --]

Hi Yehuda,

We experienced a test failure for radosgw suite on giant
backports[1], the test suite in question which failed was the bucket
quota test,

```
rgw/multifs/{clusters/fixed-2.yaml frontend/civetweb.yaml fs/xfs.yaml
rgw_pool_type/replicated.yaml tasks/rgw_bucket_quota.yaml}
```

failure logs[2] :
http://pulpito.ceph.com/loic-2015-03-27_08:40:06-rgw-giant---basic-multi/824286/

On a rerun of the same test[3], the suite came out green. Though there
wasn't anything to indicate a network failure in the first run, the
primary difference I could find between the 2 runs was that, in the
failed, run we could find this line at start:

2015-03-27T15:31:24.082 INFO:tasks.rgw.client.0.mira012.stdout:2015-03-27 15:31:24.082177
7f1bc0ff9700 -1 failed to list objects pool_iterate returned r=-2

The only patch that seemed to have made into giant that affected quotas
were https://github.com/ceph/ceph/pull/3580 (related to setting quotas
to negative values), though the test run had failed much before setting
negative quotas.

What are your thoughts on this?

[1] http://tracker.ceph.com/issues/11153#rgw
[2] http://pulpito.ceph.com/loic-2015-03-27_08:40:06-rgw-giant---basic-multi/824286/
[3] http://pulpito.ceph.com/abhi-2015-03-29_17:16:17-rgw-giant---basic-multi/

-- 
Abhishek

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 489 bytes --]