>  That said, I think Gregory may actually be recompiling xrootd himself.
> Gregory?

No, on that server I used 'xrootd-20050728-0310.i386_linux24.tgz' the 
provided binary tarball for RH7.2 so if debugging was there I should have 

> Per chance, did the executable change while xrootd was running? Based on
> the core file, gdb says that no threads have been started yet which, of
> course, doesn't jive with the log file. While it is possible that a
> massive core curruption occured, that seems highly unlikely.

I can see from the logfile I was running 20050328-0656_dbg until the crash 
but the day before had changed a symbolic link to the 20050728-0310 xrootd 
package to be used in prevision for upgrade (but did not yet restart 
xrd/olbd then). So it looks like I am the one responsible for this 

By the way... from the name it looks like I have the debugger (or does the 
_dbg suffix mean something else)?

> On a second note. Based on your client load, the number of allowed file
> descriptors is set too low (i.e., 1024). You should up the limit to at
> least 4096.

This is something I asked the admins a week ago so it should be increased 
a day or another...


-- Gregory

> On Mon, 10 Oct 2005, Gregory Schott wrote:
>> Hello,
>> On Saturday, xrootd crashed on one server without apparent reason. Olbd
>> was still running. The core file says:
>> Core was generated by `/home/xrootd/software/current/bin/xrootd -p 1094 -l /tmp/f01-016-107.xrdlog -c'.
>> Program terminated with signal 11, Segmentation fault.
>> #0  0x00000000 in ?? ()
>> (gdb) backtrace
>> #0  0x00000000 in ?? ()
>> #1  0x0806f148 in XrdPollPoll::doDetach ()
>> #2  0x0806a83b in XrdLink::XrdLink ()
>> #3  0x08070169 in XrdProtocol_Select::getProtocol ()
>> #4  0x0806f4d8 in XrdPollPoll::dqLink ()
>> #5  0x0807b21b in XrdOucLogger::XrdOucLogger ()
>> #6  0x40033fef in ?? ()
>> I put all useful files at SLAC:
>> /tmp/f01-016-107.xrdlog
>> xrootd-20050728-0310.i386_linux24.tgz
>> core.1424
>> config/
>> in ~schott/scratch/xrootd
>> Regards,
>> -- Gregory