[rudder-users] CPU load on the server

Nicolas Charles nicolas.charles at normation.com
Wed Apr 17 15:22:02 CEST 2013


On 17/04/2013 15:12, Olivier Desport wrote:
>
>
>
>
> Le 17/04/2013 15:04, Nicolas Charles a écrit :
>> On 17/04/2013 14:49, Olivier Desport wrote:
>>>
>>>
>>>
>>>
>>> Le 17/04/2013 14:30, Nicolas Charles a écrit :
>>>> On 17/04/2013 14:24, Olivier Desport wrote:
>>>>>
>>>>>
>>>>> The command still runs and the outup hangs on this last line. 
>>>>
>>>> Do you have free space left on the /var partition ?
>>>
>>> Yes, several Gb.
>>>
>>>> If you run the command
>>>> ls -hl /var/rudder/cfengine-community/state
>>>>
>>>> do you have any surprisingly large file ? Like a cf_lock.db over 50 
>>>> MB ? If so, you can safely remove this file, and check if the agent 
>>>> behaves better.
>>>>
>>>> Nicolas
>>>>
>>> You're right. The size of cf_lock.db was 307Mb ! I removed this file 
>>> and restarted rudder-server-root. Now the behaviour of cf-agent is 
>>> correct.
>>> Do I have to periodically suppress this file or can Rudder clean it ?
>>>
>>>
>> We have a mechanism on the clients to clean it when it gets 
>> corrupted, but I think we may have overlooked the server has it's 
>> much more uncommon :/
>> Did you stop the server ungracefully, or killed -9 cf-agent ?
> I've first tried /etc/init.d/rudder agent restart but it doesn't work. 
> So I've restarted the server :
>
> /etc/init.d/rudder-server-root restart
Ok, so it might have gotten corrupted by itself. It happens with BDB :( 
We are working on it.

Thank you !
Nicolas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.rudder-project.org/pipermail/rudder-users/attachments/20130417/a24b7c59/attachment.html>


More information about the rudder-users mailing list