 |
 |
Thanks Serge. OK, so no core files. Indeed, we really should nable core files all the time, even in production. Andy On Fri, 7 Aug 2015, Serge Monkewitz wrote: > Some extra info (relevant output of dmesg) below: > > ccqserv108: > > [Aug 7 22:44] xrootd[8508]: segfault at 36 ip 00007fd5b19cfc10 sp 00007fd589df8778 error 4 in libpthread-2.17.so[7fd5b19c6000+16000] > [Aug 8 00:35] xrootd[15959]: segfault at 36 ip 00007fb74b30fc10 sp 00007fb739582778 error 4 in libpthread-2.17.so[7fb74b306000+16000] > > Error code 4 means an invalid read from user-land. > > ccqserv124: > > [Aug 7 22:44] traps: xrootd[20312] general protection ip:7f98b47dac10 sp:7f98a2e72778 error:0 in libpthread-2.17.so[7f98b47d1000+16000] > [Aug 8 00:35] xrootd[30241]: segfault at 161 ip 0000000000000161 sp 00007f25a67be798 error 14 in xrootd[400000+13000] > > error 0 is an invalid kernel read. > error 14 is an invalid user-land write, with reserved bit set (no idea what it means for the reserved bit to be set). > > I˙˙m not sure how useful this is without stacktraces. I would suggest adding "ulimit -c unlimited˙˙ to the xrootd startup scripts (/qserv/run-jgates/etc/init.d/xrootd on all nodes) so that you get core files next time. > ######################################################################## Use REPLY-ALL to reply to list To unsubscribe from the QSERV-L list, click the following link: https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1