Print

Print


Hi Geonmo,

Running out of space in the log file may cause a hang like you see but it 
is more likely you ran out of file descriptors. The log will show that 
problem for the latter case. For the lg case, you can attach gdb to the 
running process and issue the following commands:

set logging on
set logging redirect on
set pagination off
thread apply all bt full

This will create a log file named gdb.txt. So, send both the log file and 
the gdb.txt file to me (it's not a good idea to post these to xrootd-l for 
privacy reasons).

Andy

On Mon, 30 Sep 2019, "Geonmo Ryu" wrote:

> Dear XRootD experts,Recently we found the xrootd service to behave abnormally. So, I send you an email to inquire about this.About a month ago we upgraded the versions of the XRootD and dCache programs and set up the XRootD CMS AAA access to the dCache XRootD Door using the proxy feature of XRootD.As a result, the service worked very well and reliably against massive data access requests.(The xrootd process has a CPU load of 200-300% and 3.9G of RSS in use. [It's been 3 hours since restarting.] It doesn't measure correctly, but it seems to be about 20-30 requests per second. As mentioned earlier, the process is acting as a proxy service.)However, after one or two weeks of starting up the XRootD service, I found that the XRootD server had a problem.The problem is that the symptoms were not always the same.In some cases, the service itself hangs, and when we connect using xrdfs, it did not show a prompt after ls command.In other cases, the file can be listed, but when 
 I try to access it, it was failed due to a zombie state of file.At this time, log files are not properly collected, so it is difficult to confirm the problem.Please tell me how to set up the debugging option, and I will send you a log if the problem recurs after setting.I don't know if it is the cause, but when this problem occurred, there was not enough space for log in the /var/log directory. However, since the problem did not occur as soon as capacity ran out, it was not possible to determine whether the cause was insufficient space.(The service was stopped for about 17 hours after it ran out of capacity and was unable to log.)Regards.
>
> --------------------------------------------------------------------------------------------------
> Geonmo Ryu / ???
>
> Korea Institute of Science and Technology Information (KISTI)
> Global Science Experimental Data Hub Center (GSDC)
> 245 Daehak-ro, Yuseong-gu, Daejeon, 305-806, Republic of Korea
> Tel :  +82-42-869-1639, +82-10-4337-9423
> Mail : [log in to unmask] / [log in to unmask]
> -------------------------------------------------------------------------------------------------- 
>
>
>  
>
>
>
> ########################################################################
> 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