This past weekend I upgraded 22 multiprocessor Enterprise Redhat Linux
servers (ver 5.4, and 5.5) from BIND 9.6-ESV-R5-P1 to 9.6-ESV-R7-P1. Six
of the servers began to BIND exit after having run for 8 to 12 minutes.
All 22 server configured with: ./configure --prefix=/usr --sysconfdir=/etc
--localstatedir=/var --enable-threads
The 6 servers that were failing were running different version of Redhat
(5.4, and 5.5) These are the same versions that the other 16 servers are
running an they have no problems.
The 6 servers that failed were those that are used to resolve DNS names
FROM the internet, they point at internet roots, the 16 servers that did
NOT fail are internal caching resolvers that point at internal root
servers.
The following was discovered in the log file.
Jun 10 00:39:14 NS1 named[9631]: general: critical: random.c:106:
REQUIRE(jitter < max) failed
Jun 10 00:39:14 NS1 named[9631]: general: critical: exiting (due to
assertion failure)
The only difference I can see in random.c are lines 2 and 18 but both are
commented out so should have no impact. Does anyone have any idea what
could have changed in the way 9.6-ESV-R7-P1 uses random.c or what values
it passes to random.c? Seems like 9.6-ESV-R7-P1 is passing random.c some
jitter value that it does not like so it exits. Any chance this could be
related to the name servers is the internet root hints file? I noticed
that I was using an old version of the root name servers list.
The 6 failing servers we downgraded back to BIND 9.6-ESV-R5-P1 and are
running fine now.
David A. Evans
Enterprise IP/DNS Management
Network Infrastructure Tools and Services
evans_davi...@cat.com
_______________________________________________
Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe
from this list
bind-users mailing list
bind-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/bind-users
