Bug #15597
closedrgw: when upgrading to jewel, a default zone setup is is not upgraded correctly
0%
Description
- hammer / infernalis
- create a default setup (don't set up zone)
- create user + bucket
- stop + upgrade to jewel
- radosgw-admin user info --uid=<uid>
Updated by Yehuda Sadeh almost 8 years ago
- Priority changed from Normal to Urgent
Updated by Abhishek Lekshmanan almost 8 years ago
Master PR: https://github.com/ceph/ceph/pull/8766
Updated by Nathan Cutler almost 8 years ago
- Status changed from New to Fix Under Review
Updated by Herve Rousseau almost 8 years ago
Got also hit by this on a brand new Jewel cluster (so nothing to upgrade).
I have uploaded radosgw logs with debug_rgw set to 20 at
ceph-post-file: 0161c5aa-a161-4c9c-94e5-89b4b0644141
If you have a gitbuilder rpm I'd be happy to test it !
Updated by Yehuda Sadeh almost 8 years ago
Herve, maybe you see a different issue? This bug is specifically about failing to upgrade. Looking at your log it seems that your radosgw is processing requests. What are you actually seeing? Can you also run it with 'debug ms = 1'? Thanks
Updated by Herve Rousseau almost 8 years ago
Indeed, it could be a different problem. As the symptom is very similar I thought they would be somewhat related.
After some more testing, I confirm that I can do all operations, it's just very annoying to get these errors each time I instantiate radosgw-admin
and very slow as it tries to list/create/set the zone.
The request that you see is actually just some health check script that parses the HTTP return code (of GET /), no S3 operation here.
2016-04-28 21:50:53.230504 7fb087875a40 0 RGWZoneParams::create(): error creating default zone params: (17) File exists
Here's the radosgw log with a clean cluster and debug_ms=1 in addition to debug_rgw:
ceph-post-file: 3bf4fcf2-8a3b-425c-9c30-e6ad01743f73
Updated by Abhishek Lekshmanan almost 8 years ago
Herve Rousseau wrote:
Indeed, it could be a different problem. As the symptom is very similar I thought they would be somewhat related.
After some more testing, I confirm that I can do all operations, it's just very annoying to get these errors each time I instantiate
radosgw-admin
and very slow as it tries to list/create/set the zone.The request that you see is actually just some health check script that parses the HTTP return code (of GET /), no S3 operation here.
2016-04-28 21:50:53.230504 7fb087875a40 0 RGWZoneParams::create(): error creating default zone params: (17) File exists
We could probably increase the verbosity for error messages like these so that they dont get printed for the regular case so that it is less confusing for the user?
[...]
Here's the radosgw log with a clean cluster and debug_ms=1 in addition to debug_rgw:
[...]
Updated by Herve Rousseau almost 8 years ago
Abhishek Lekshmanan wrote:
We could probably increase the verbosity for error messages like these so that they dont get printed for the regular case so that it is less confusing for the user?
Hi,
I think more work is needed than just changing the loglevel of the message, the cluster has no buckets and still it takes more than 1 minute to get the empty list back:
$ time radosgw-admin --id radosgw bucket list 2016-04-29 16:53:01.012824 7f90e224ba40 0 RGWZoneParams::create(): error creating default zone params: (17) File exists [] real 1m2.578s user 0m0.651s sys 0m0.415s
Updated by Yehuda Sadeh almost 8 years ago
- Status changed from Fix Under Review to Pending Backport
Updated by Nathan Cutler almost 8 years ago
- Copied to Backport #15803: jewel: rgw: when upgrading to jewel, a default zone setup is is not upgraded correctly added
Updated by Abhishek Lekshmanan almost 8 years ago
@Herve Rosseau,
can you open a different issue to track the radosgw-admin operation slowness, are you seeing this only for bucket listing or for user ops etc as well?
Updated by Nathan Cutler almost 8 years ago
- Status changed from Pending Backport to Resolved