Awesome thanks :))
I have found it
When I list bucket a bucket is exists but when I get bucket stats for that
bucket it again fails with -2!
This will happens:
`couldn't put binfo cache entry, might have raced with data changes`
And rapidly gets this error for each OSD that mon trying to get info
```
osd_op(unknown.0.0:1 49.13a
49:5cdd94e7:::.dir.ac7cb16a-c4d4-4e8e-a6ad-97898dc5af11.108925803.3.0:head
[call rgw.bucket_list] snapc 0=[] ondisk+read+known_if_redirected e214222)
v8 -- 0x5622f89a6700 con 0x5622f89a3a60
conn(0x5622f89a97e0 0x5622f89abcc0 unknown :-1 s=NONE pgs=0 cs=0 l=1 rx=0
tx=0).connect
```
Until I get `error getting bucket stats`
On Fri, May 8, 2020 at 6:45 PM Casey Bodley <cbodley(a)redhat.com> wrote:
You can add '--debug-rgw 20 --debug-ms 1' to
this radosgw-admin
command. You don't have to change log levels for radosgws.
On Fri, May 8, 2020 at 10:04 AM Seena Fallah <seenafallah(a)gmail.com>
wrote:
So how can I debug that which bucket is getting error? I will still get
that error
without any further information!
If you mean that I change the debug level to 20
and run this command to
see what’s wrong I think it’s getting hard to debug it! I’m
in a production
mode and I should change config reload services and also look forward a
long and heavy log file in radosgw to find out what’s wrong! This gets more
cost and the error says which bucket is getting this error so I will try
just on that bucket and trace that bucket with debug 20.
If I set debug to 20 now there will be a heavy
logs comes from radosgw
for sharding and others that gets mess!
Don’t you agree with it? It really hard to change log level in
production mode :(
On Fri, May 8, 2020 at 6:28 PM Casey Bodley <cbodley(a)redhat.com> wrote:
>
> On Fri, May 8, 2020 at 9:52 AM Seena Fallah <seenafallah(a)gmail.com>
wrote:
> >
> > Im just calling an admin api for getting all usage for all buckets. I
don’t specify any buckets there! And also sometimes getting bucket stats
from radosgw-admin that if filtered by uid that is exists and against I
don’t specify any bucket!
>
> The `radosgw-admin bucket stats` command (without additional arguments
> like --uid or --bucket) will list all buckets the same way that the
> admin api does.
>
> >
> > On Fri, May 8, 2020 at 6:18 PM Casey Bodley <cbodley(a)redhat.com>
wrote:
> >>
> >> The `radosgw-admin bucket stats` will run the same code as this admin
> >> api, so you can debug the command itself without touching your
> >> gateways.
> >>
> >> On Fri, May 8, 2020 at 9:28 AM Seena Fallah <seenafallah(a)gmail.com>
wrote:
> >> >
> >> > I’m in a production mode and I can’t go on debug 20 always. This
error is useless by it self, if it contains bucket name it would be so
helpful than it just says error something that isn’t available (-2).
> >> >
> >> > On Fri, May 8, 2020 at 5:55 PM Casey Bodley <cbodley(a)redhat.com>
wrote:
> >> >>
> >> >> On Fri, May 8, 2020 at 6:45 AM Seena Fallah <
seenafallah(a)gmail.com> wrote:
> >> >> >
> >> >> > Hi all.
> >> >> >
> >> >> > I'm getting lots of error getting bucket stats from my rgw
instances
> >> >> > and I can't figure
out what's going wrong!
> >> >>
> >> >> can you enable debug_rgw=20 and debug_ms=1? that output will
include
> >> >> the relevant bucket and
bucket instance names. if you suspect bugs
> >> >> here, please open a tracker issue with the logs!
> >> >>
> >> >> > Can we add bucket name to this error to see whats happening in
the cluster?
> >> >> >
https://github.com/ceph/ceph/pull/34653
> >> >> >
> >> >> > Thanks.
> >> >> > _______________________________________________
> >> >> > Dev mailing list -- dev(a)ceph.io
> >> >> > To unsubscribe send an email to dev-leave(a)ceph.io
> >> >> >
> >> >>
> >>
>