On Tue, May 12, 2015 at 5:54 AM, Kenneth Waegeman
<kenneth.waegeman@xxxxxxxx> wrote:
>
>
> On 04/30/2015 07:50 PM, Gregory Farnum wrote:
>>
>> On Thu, Apr 30, 2015 at 2:03 AM, Kenneth Waegeman
>> <kenneth.waegeman@xxxxxxxx> wrote:
>>>
>>> So the cache is empty, but I get warning when I check the health:
>>> health HEALTH_WARN
>>> mds0: Client cephtst.cubone.os failing to respond to cache
>>> pressure
>>>
>>> Someone an idea what is happening here?
>>
>>
>> This means that the MDS sent the client a request to drop some cached
>> inodes/dentries and it isn't. This could mean that the client is too
>> old to respond correctly, or that you've actually got so many files
>> open that the client can't drop any of the requested items.
>
>
> Can we increase the maximum allowed number of inodes for the client(s) ? I
> thought there was a setting for that, but I don't seem to find it anymore..
It's not a fixed number per-client — if the MDS cache is oversized, it
will just ask for clients to return state proportional to how much
it's oversized by. If the MDS gets a larger cache ("mds cache size",
default 100000) then clients can use more files at once.
-Greg
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
