As part of security testing and to keep track of internal machine I times I
run rup. All 4 of my RH70 boxes respond. I have started rstatd on the
beta3 boxes and
it starts just fine but immediately stops w/o an error logged.
This defect is considered MUST-FIX for Florence Gold release
Can you start the rstatd using strace(1) and attach the output to the bug
This is a SEGV during a call to daemon(3).
Bob, can you please be more verbose about why you think it segfaults in daemon?
Have you received any strace and/or debugged this a little bit?
I don't see how a program could crash in daemon (unless it is linked
against -lpthread and for some reason thread local register is broken
but rpc.rstatd is not linked against -lpthread).
Jakub, this was my mistake. I saw the segv after the daemonize call and assumed
that it was happening there. But, a little more investigation shows that the
child process is segfaulting long after the return from fork.
I will try and fix this quickly, and if I can't will reclassify as rusers bug.
/proc/stat layout has changed in 2.4 kernels. Patch attached.
Created attachment 9961 [details]
makes rstat_proc.c live happily with 2.4 /proc/stat
Bob, patch will be pulled into the next rusers build. Thanks!