On Mar 19 2013, Stephen Wood wrote:
We are measuring the performance of our bind9 cache & forward servers.
Under "++ Resolver Statistics ++", we see a high number of "queries aborted
due to quota".
These are queries discarded because the limit on outstanding recursive
queries ("recursive-clients" in "options") has been exceeded. You may
need to increase the default of 1000, or there may be some other problem.
Use "rndc status" and check how the "recursive-clients" values are
behaving. If the queues are regularly long, snapshot them by using
"rndc recursing" and see what you can make of the ones that are taking
longest.
I can't find information regarding this number in the documentation.
Overall, the SERVFAIL numbers are low. Are aborted queries simply retried?
When the limit is exceeded, the oldest outstanding query is discarded,
without giving any response to the client. The client may retry, of course.
Note that BIND gives up on recursive queries that have not completed
after 30 seconds, in any case.
Do I need to make any kernel level changes, like allowed sockets or open
files? How do I troubleshoot this message?
If you have to increase "recursive-clients" a lot, you may also need to
increase named's -S option, or it may run out of sockets before query
queue slots.
I have to say I am a bit suspicious of your reference to "forward servers".
There are all sorts of possible misconfigurations using forwarders that
might provoke problems of this sort.
--
Chris Thompson
Email: c...@cam.ac.uk
_______________________________________________
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
