jim at rfc1035.com
Mon Mar 20 21:46:07 UTC 2000
>> It seems like once a week, named dies: pid82 (named), uid0:
>> exited on signal 11
>> and today I got a core dump. The TTL is set to 86400 (default,
>> I think), but it shouldn't crap out like this, should it? It's
>> only been doing this for the last three weeks, or so. I restart
>> the process and everything runs fine for the week.
>> Appreciate any info...
It would have helped if you'd give information like the OS platform or
what version of name server code you were running and, more
importantly,what messages were printed in the logs prior to the name
The name server does not make a habit of dying as a general
rule. Signal number 11 is a segmentation violation. This is usually
generated by an illegal memory reference - like following an incorrect
pointer. In general the name server doesn't do this. Can you correlate
the crash to some other DNS activity - like something sending a
specific query or malformed packet to the name server? Or could it
just be that your name server is hitting an OS default on the amount
of VM that a process can use?
If you're not running up to date name server code, upgrade. I doubt if
anyone cares if there *might* be a pointer bug in obsolete DNS code.
More information about the bind-users