Hi Tommaso,
When this happens it could be a combination of two things -- too much
memory being used by the network stack (Linux has a limit) which means
it's overloaded, not enough memory left for direct filesystem access (file
system overloaded). As for threads, there would be other messages in the
log. 2033 connection is not that much (we sometimes run almost an order of
magnitude more). If you send me the log for the day I may be able to get
additional information for you. It simply looks like you managed to
overload the server. I though you were using Brian's throttling plug-in to
avoid that. I guess not.
Andy
On Wed, 9 Jul 2014, Tommaso Boccali wrote:
> Ciao, I have a server under very heavy load (1 GB/s incoming, 200 MB/s
> outgoing - these are bytes, not bits!), which is failing all the xrootd
> file serving, with stuff like
>
> 140709 16:22:20 7453 XrdLink: Unable to send file to
> cmsplt00.41552:[log in to unmask]; operation canceled
> 140709 16:22:20 7453 XrootdXeq: cmsplt00.41552:[log in to unmask] disc
> 0:05:21 (sendfile failure)
>
> I see no other real problems on that ... any hint on what can be the
> problem?
> FS? network saturated?
>
> I really see tons of rootd connections:
>
> [root@stormgf1 cms]# netstat|grep rootd|wc
> 2033 12203 180956
>
> this is 5 min after a restart, and they are still increasing ....
>
> the other thing I do not get is why the other server (stormgf2.pi.infn.it)
> is getting instead < 1/2 of the files
>
>
> can you confirm the error is just due to high traffic and NOT to (for
> example) max threads reached or so?
>
> tom
>
> --
> Tommaso Boccali
> INFN Pisa
>
> ########################################################################
> Use REPLY-ALL to reply to list
>
> To unsubscribe from the XROOTD-L list, click the following link:
> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
|