Statistics / Nagios

Statistics / Nagios

[Stefan Priebe]

Hello list,

are there any useful statistics for ceph? ceph -s shows just the status 
but i would like to record something which gives me an idea about the 
ceph load or usage.

Greets,
Stefan

Re: Statistics / Nagios

[Gregory Farnum]

On Sunday, November 25, 2012 at 10:26 AM, Stefan Priebe wrote:
> Hello list,
> 
> are there any useful statistics for ceph? ceph -s shows just the status 
> but i would like to record something which gives me an idea about the 
> ceph load or usage.

Each daemon has an admin socket which you can use to extract "perfcounters" (performance counters). It's not well-documented at all right now, unfortunately.
"ceph --admin-daemon /path/to/admin/socket help"
"ceph --admin-daemon /path/to/admin/socket perfcounters_dump"

"ceph --admin-daemon /path/to/admin/socket perfcounters_schema"

should get you started.
http://ceph.com/docs/master/radosgw/troubleshooting has some examples of troubleshooting the gateway with them.
-Greg

Re: Statistics / Nagios

[Stefan Priebe]

Hi Greg,

Am 25.11.2012 20:08, schrieb Gregory Farnum:
> On Sunday, November 25, 2012 at 10:26 AM, Stefan Priebe wrote:
>> Hello list,
>>
>> are there any useful statistics for ceph? ceph -s shows just the status
>> but i would like to record something which gives me an idea about the
>> ceph load or usage.
>
> Each daemon has an admin socket which you can use to extract "perfcounters" (performance counters). It's not well-documented at all right now, unfortunately.
> "ceph --admin-daemon /path/to/admin/socket help"
> "ceph --admin-daemon /path/to/admin/socket perfcounters_dump"
>
> "ceph --admin-daemon /path/to/admin/socket perfcounters_schema"
>
> should get you started.
> http://ceph.com/docs/master/radosgw/troubleshooting has some examples of troubleshooting the gateway with them.

thanks! I don't use radosgw. Just ceph-mon and ceph-osd with rbd block 
devices.

So i have to query each OSD on their own?

Right now i have
[global]
   admin socket=/var/run/ceph.sock

but that gives:
# ceph --admin-daemon /var/run/ceph.sock perfcounters_schema
{}

Greets,
Stefan

Re: Statistics / Nagios

[Dan Mick]

Right; you'd want to make separate socket paths for each daemon 
(something like admin socket = /var/run/$cluster-$name.sock should do,
although note that the default is already /var/run/ceph/$cluster-$name.asok)

On 11/25/2012 11:32 AM, Stefan Priebe wrote:
> Hi Greg,
>
> Am 25.11.2012 20:08, schrieb Gregory Farnum:
>> On Sunday, November 25, 2012 at 10:26 AM, Stefan Priebe wrote:
>>> Hello list,
>>>
>>> are there any useful statistics for ceph? ceph -s shows just the status
>>> but i would like to record something which gives me an idea about the
>>> ceph load or usage.
>>
>> Each daemon has an admin socket which you can use to extract
>> "perfcounters" (performance counters). It's not well-documented at all
>> right now, unfortunately.
>> "ceph --admin-daemon /path/to/admin/socket help"
>> "ceph --admin-daemon /path/to/admin/socket perfcounters_dump"
>>
>> "ceph --admin-daemon /path/to/admin/socket perfcounters_schema"
>>
>> should get you started.
>> http://ceph.com/docs/master/radosgw/troubleshooting has some examples
>> of troubleshooting the gateway with them.
>
> thanks! I don't use radosgw. Just ceph-mon and ceph-osd with rbd block
> devices.
>
> So i have to query each OSD on their own?
>
> Right now i have
> [global]
>    admin socket=/var/run/ceph.sock
>
> but that gives:
> # ceph --admin-daemon /var/run/ceph.sock perfcounters_schema
> {}
>
> Greets,
> Stefan
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Statistics / Nagios

[Gregory Farnum]

On Sun, Nov 25, 2012 at 11:32 AM, Stefan Priebe <s.priebe@profihost.ag> wrote:
> Hi Greg,
>
> Am 25.11.2012 20:08, schrieb Gregory Farnum:
>
>> On Sunday, November 25, 2012 at 10:26 AM, Stefan Priebe wrote:
>>>
>>> Hello list,
>>>
>>> are there any useful statistics for ceph? ceph -s shows just the status
>>> but i would like to record something which gives me an idea about the
>>> ceph load or usage.
>>
>>
>> Each daemon has an admin socket which you can use to extract
>> "perfcounters" (performance counters). It's not well-documented at all right
>> now, unfortunately.
>> "ceph --admin-daemon /path/to/admin/socket help"
>> "ceph --admin-daemon /path/to/admin/socket perfcounters_dump"
>>
>> "ceph --admin-daemon /path/to/admin/socket perfcounters_schema"
>>
>> should get you started.
>> http://ceph.com/docs/master/radosgw/troubleshooting has some examples of
>> troubleshooting the gateway with them.
>
>
> thanks! I don't use radosgw. Just ceph-mon and ceph-osd with rbd block
> devices.
>
> So i have to query each OSD on their own?

That's how it stands at present, yep.

> Right now i have
> [global]
>   admin socket=/var/run/ceph.sock
>
> but that gives:
> # ceph --admin-daemon /var/run/ceph.sock perfcounters_schema
> {}

Hrm, I'm getting a full schema out of it when I test locally. I
suspect that maybe you're trying to query the admin tool's socket when
you do that — everybody, including the client tools, will export one
if it's set for them. Try using the $name metavariable as part of the
path.
-Greg
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html