[ 
https://issues.apache.org/jira/browse/TS-1804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13623981#comment-13623981
 ] 

Leif Hedstrom commented on TS-1804:
-----------------------------------

So first, you should always set net.connections_throttle to what you expect it 
to be. We guarantee that the traffic_server process gets a least that many 
max-files via setrlimit(). It never makes sense to give ATS more max-files than 
number of connections anyways.

Secondly, as far as I understand, /etc/security/limits.conf only applies to 
users logged in via PAM. And that certainly explains your behavior (processes 
started via the bootup system does not use limits.conf). From my limits.conf 
file:

{code}
# /etc/security/limits.conf
#
#This file sets the resource limits for the users logged in via PAM.
#It does not affect resource limits of the system services.
{code}


                
> fds_limit improperly set upon server reboot + ats restart
> ---------------------------------------------------------
>
>                 Key: TS-1804
>                 URL: https://issues.apache.org/jira/browse/TS-1804
>             Project: Traffic Server
>          Issue Type: Bug
>          Components: Core
>            Reporter: David Carlin
>
> proxy.config.net.connections_throttle = 70000 in records.config.
> /etc/security/limits.conf:
> * - nofile 500000
> When I reboot a server and start ATS, every time ATS will set fds_limit to 
> value of proxy.config.net.connections_throttle instead of the per-user max 
> set in limits.conf - it is my understanding it should pick limits.conf value 
> if its higher.
> Subsequent restarts of ATS will always pick the limits.conf value, the 
> problem only occurs when starting ATS after a server reboot.
> This was causing us to experience the error 'too many open file descriptors, 
> emergency throttling' on random hosts that had too low a value for fds_limit.
> The workaround is to set proxy.config.net.connections_throttle higher or 
> restart ATS twice upon server startup.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to