BIND 8.2.1 Hangs - NDC fails

Adam D. Burstein adam at brodie.two-four.com
Wed Sep 8 04:48:20 UTC 1999


I have seen several posts related to this problem, but I haven't seen any
fixes.

We've recently upgraded to BIND 8.2.1, and it sporadically hangs for
periods of up to 60 minutes (the longest I've seen as yet).  Out of
nowhere, the name server will begin to respond again as though nothing
happened.

Although we have logging enabled, there is usually nothing in the log file
to indicate a problem.  On occasion, we will see something like:

ns_req: sendto([IP ADDRESS REMOVED].53): Connection refused
(where IP ADRESS REMOVED represents the real IP address)

or

stream_getlen([IP ADDRESS REMOVED].2896): Broken pipe

We are also getting a lot of unauthorized updates (which I would like to
prevent, if someone could please advise me on how to deny incoming
non-query traffic -- is that even possible?).

When the server hangs, I try to reload the server using ndc reload (as
opposed to signals), but the server does not respond to the request.
Sending a SIGHUP causes the server to initiate its reload. 

We experience this problem predominantly on primary name servers.  Despite
all name servers having identical hardware and software configurations,
the problem does not appear to surface on all machines.

Could this be related to the unauthorized updates/AXFRs tying up channels,
sockets, etc.?  Just a guess. 

Thanks in advance...
Adam

-- 
Adam D. Burstein
adam at brodie.two-four.com



More information about the bind-users mailing list