On Wed, 19 Aug 1998, Smeriglio,Luca,VEVEY,FC-SIL/INF. wrote:
> Hello list !
> Just a small one...
> netmon (and ovtopmd) is stopping everyday on my nv6k 5.0, because
> of ovtopmd losing connection with trapd.
> It restarts correctly when I do it manually (ovstart netmon). It's just
> really annoying.
> (and, of course, no polling during that time...)
> Any ideas ?
The key here would be to find out why trapd is going away. Does it leave a core
file? If so, you will need to contact support. They will need you to send in
the core, as well as the output of /usr/OV/service/readcore (cd to the directory
containing the core file and run the command without arguments).
Most of the deamons will leave their core files in /usr/OV/PD/<deamon name>, but
I think trapd is an exception. To find out where the core file is, you need to
do some investigating. Find the most recent errorlog entry that contains the
message for the core dump (errpt -aJ CORE_DUMP). (The program that cored is the
last line of the entry). Above the last line there are two pieces of info you
need to be able to locate where the core is:
FILE SYSTEM SERIAL NUMBER and INODE NUMBER. Write these down.
print mount | crash
Look for the line that has the number indicated by the FILE SYSTEM
SERIAL NUMBER from errpt in the NUMBER column of the output. The mount
point is indicated by the last column. Then do:
find <mount point indicated by above> -inum <INODE NUMBER from errpt> -xdev
This will return the directory name that contains the core file.
Now you will have to make sure that you take care of 3 things before
you can get a full core dump, because a partial core dump may not have
enough data for support to tell what is going on.
1. chdev -l sys0 -a fullcore=true (if it isn't already set)
2. edit /etc/security/limits and make the following change the default stanza:
core = 65536
(Make sure that root doesn't have a core stanza that is less than this)
3. If you had to make the last change, you will have to log out and back in
and then restart trapd (at least - you may have to stop all deamons, including
ovspmd, and restart them).
Ray Schafer The Kernel Group http://www.tkg.com
schafer AT tkg DOT com +1 212 880 6444 Pager: (800) 532-4722