BIND 9.10 dlz(BDB): slow query times

ShanyiWan wsy at 114.com.cn
Mon Nov 10 03:52:51 UTC 2014


 
Tested and found that the Buffer is full(udp Recv-Q full)!
A large number of requests generate a lot of logs, resulting in heavy disk IO.
So modify logging:
channel bind_log { 
file "/var/log/named/bind.log" versions 3 size 50m; 
severity warning; 
#severity info; 
print-time yes; 
print-severity yes; 
print-category yes; 
}; 
category default{ 
bind_log; 
};

And now,works fine!


ShanyiWan
 
From: Evan Hunt
Date: 2014-11-07 01:22
To: ShanyiWan
CC: bind-users
Subject: Re: BIND 9.10 dlz(BDB): slow query times
On Fri, Nov 07, 2014 at 12:27:55AM +0800, ShanyiWan wrote:
> 100000+ domains, 500000 records,
> 
> Most of the query time in 1-3 seconds, is this normal?
 
DLZ is slow, but it's not that slow.  I would guess the delay you're
seeing is due to the time it takes to retrieve data from the berkeley
DB.  How long does it take to query the database directly?
 
Turning on minimal-responses reduces the number of database lookups
required for each query, which might help if the database is slow.  There
may also be a bdb mailing list that can recommend database optimizations.
 
-- 
Evan Hunt -- each at isc.org
Internet Systems Consortium, Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20141110/dafafb66/attachment.html>


More information about the bind-users mailing list