We've got the same problem here (FWIW, etch on amd64). A bit of digging
traces it back to munin's nfsd plugin -- the beginning of the segfaults
coincides with that box being replaced as our NFS server.
The plugin is returning garbage, because the file it scrapes doesn't
exist any more:
$ nc <elided> 4949
# munin node at <elided>
fetch nfsd
getattr.value setattr.value lookup.value access.value readlink.value
read.value write.value create.value mkdir.value symlink.value
mknod.value remove.value rmdir.value rename.value link.value
readdir.value readdirplus.value fsstat.value fsinfo.value pathconf.value
commit.value
.
This is easily fixable by deleting the plugin symlink, but it would be
nice if munin-update didn't fail so badly when faced with bad data.
The problem on the RRDtool side looks to be a bit harder to identify. A
quick run through gdb gives the same backtrace as Ferenc Wagner reported
earlier. It seems that rrd_get_context() is returning a garbage
pointer. I'll take a closer look at where that pointer is coming from
tomorrow, if I get the time.
--Matt
--
Matthew Boyle
Junior Systems Administrator
DecisionSoft Limited http://www.decisionsoft.com
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
