[Spacewalk-list] repo sync crashing

Michael Mraka michael.mraka at redhat.com
Thu Feb 10 20:01:21 UTC 2011


Sander Grendelman wrote:
% This looks like a "too many open files" error ( 1014 is suspiciously
% close to 1024).
% 
% On Thu, Feb 10, 2011 at 06:58:47AM -0500, Kenneth Ratliff wrote:
% > I've been noticing that my repo syncs haven't been completing an actual full sync. I had been syncing them through the web GUI, but tonight I tried it via the commandline to see if I could see what was happening. It usually dies after transferring 1013 packages. 
% > 
% > And sure enough, doing so tonight yielded the following as soon as it went to transfer the 1014th package (I'm syncing the entire EPEL repo, so this has taken a little bit) 
% > 
% > python: ./rpmio_internal.h:502: c2f: Assertion `fd && fd->magic == 0x04463138' failed.
% > ORA-24550: signal received: [si_signo=6] [si_errno=0] [si_code=-6] [si_int=4202816] [si_ptr=0x402140] [si_addr=0x4848]
% > kpedbg_dmp_stack()+297<-kpeDbgCrash()+75Cannot open /usr/lib/oracle/11.2/client/lib/libclntsh.so.11.1 for reading: errno=24
% > <-075E6CC1Cannot open /usr/lib/oracle/11.2/client/lib/libclntsh.so.11.1 for reading: errno=24
% > <-07377F6F<-gsignal()+80<-abort()+257<-__assert_fail()+251Cannot open /usr/lib/librpmio-4.4.so for reading: errno=24
% 
% Error 24 is too many open files (from the top of my head).
% 
% Try upping max open files for root (and please check if your oracle user
% has a nice high ulimit -n too).

Bingo :).

It was a leaked file descriptor. It's been fixed in nightly.
Thanks both of you for the report and the hint.

Regards,

--
Michael Mráka
Satellite Engineering, Red Hat




More information about the Spacewalk-list mailing list