Bind 9.3.4-P1 Crashing?

[David Nolan]

(sent this earlier, but it hasn't gone through for some reason....)

In the last two weeks I've seen Bind 9.3.4-P1 crash three times on my
recursive DNS servers.  In each occasion multiple redundant servers
crashed nearly simultaneously.   In each case bind has just apparently
died spontaneously with no errors logged anywhere, the servers
themselves continued to run as normal.  The first time occurred two
weeks ago at 5AM on a Saturday on two servers providing identical
virtual server addresses.  The same pair of servers died again this
morning around 6AM, and again at 10AM, but a third server providing
different service died as well that time.

These are linux machines, running 2.6.16.29.  The kernel hasn't changed
on these machines recently.  The last relevant thing that changed on
these machines should be when we upgraded to 9.3.4-P1 when it was
released.

I'm currently doing some packet collection to attempt to track this
down if/when it happens again.   I have also upgraded one of the servers to 
9.4.1-P1 to see if next time only one fails.

My questions for the list are:

Any one else seeing unexpected bind crashes in the last couple of weeks?

Are there any other debugging steps I should be taking now to provide
maximum useful data when it happens again?


-David Nolan
 Network Services
 Carnegie Mellon University