On Thursday, April 24, 2014, Georg Höllrigl <[email protected]>
wrote:

>
>  And that's exactly what it sounds like — the MDS isn't finding objects
>> that are supposed to be in the RADOS cluster.
>>
>
> I'm not sure, what I should think about that. MDS shouldn't access data
> for RADOS and vice versa?


The metadata for CephFS is all stored in RADOS in the "metadata" pool. The
MDS absolutely has to access it. :)


>
>  Anyway, glad it fixed itself, but it sounds like you've got some
>> infrastructure issues or something you need to sort out first.
>> -Greg
>> Software Engineer #42 @ http://inktank.com | http://ceph.com
>>
>>
> I think we found a reason - somehow all the memory gets used up - maybe
> some leak? So ATM it's not really fixed.


There's a page in ceph.com/docs on gathering memory statistics; you could
do that. You can also get a cache dump out of the MDS. Unfortunately we
can't do much else just now.
(You should search through the docs for those keywords; I'm on my phone
right now so I can't give you more precise instructions.)
-Greg


>
> Is there anything I could do, so that we could track down and fix this in
> future versions?
>
>
> Regards,
> Georg
> _______________________________________________
> ceph-users mailing list
> [email protected]
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>


-- 
Software Engineer #42 @ http://inktank.com | http://ceph.com
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to