[Gluster-users] Monitoring / Stats

[Vijay Bellur]

On 05/07/2015 12:03 PM, Michael Schwartzkopff wrote:
> Am Mittwoch, 6. Mai 2015, 23:34:31 schrieb Oliver:
>> Hi gluster users,
>>
>> I have a few questions about monitoring and statistics.
>>
>> We are about to install a distributed/replicated volume on multiple
>> servers for redundancy and scalability on ubuntu14.
>>
>> How can I ensure that really all files are distributed to multiple
>> servers ? Do I have any chance to see stats anywhere how many replicas
>> of a file are located on which servers or so ?
>>
>> What about adding bricks later, how can I again ensure that after a
>> rebalancing everything is spreaded over the new count of bricks/replicas ?
>>
>> We use icinga2 for monitoring, my best idea to make sure that the fs is
>> available is to write a file incl a timestamp to the fs via cron on any
>> server and to read it from another server and send some passive check
>> result to the icinga server. Any better ideas anyone maybe ? Does a best
>> practise exist therefore ?
>>
>> Also, I would love to see some insight stats in our graphite setup, I
>> already saw diamond-collectors for ceph, does anyone know about similar
>> things for gluster ?
>>
>> thanks in advance,
>>
>> Oliver
>
> What gluster really needs is a good SNMP agent to deliver fault and
> performance data. SNMP is the standard for all monitoring systems regardless
> if open source or commercial.
>

gluster-nagios [1] does send SNMP traps for alerting. More details at 
[2]. Would that be useful for you?

-Vijay

[1] http://download.gluster.org/pub/gluster/glusterfs-nagios/
[2] http://www.ovirt.org/Features/Nagios_Integration