[bind10-dev] b10-auth stops responding for several tens of seconds after receiving IXFR
JINMEI Tatuya / 神明達哉
jinmei at isc.org
Tue Mar 19 16:19:32 UTC 2013
At Tue, 19 Mar 2013 14:59:48 +0900,
Yoshitaka Aharen <aharen at jprs.co.jp> wrote:
> We've been running BIND10 1.0.0 in our test environment which retrieves
> periodic jp zone transfer into in-memory DB with SQLite3 backend.
> Retrieving IXFR consists of 2 parts: SQLite3 update by Xfrin module and
> (full) zone reload of Auth modules.
> The time required to SQLite3 DB update depends on the number of updated
> records and it takes about 2.5 minutes for 70 records of update in
> average. It seems to be bit slower, but it should be because the test
> environment is old; 2 AMD Opteron 275 processor and 8GB of memory,
> 1.5Gbps SATA 250GB HDD.
Thanks for the experiments. The update speed seems to be too slow
(taking more than 2sec for 1 update). I don't think it's due to the
machine power. There should be some significant bottleneck either in
our own code or in a way we use SQLite3.
Do you have more detailed information about which part of the update
processing takes time? Maybe you can experimentally increase the log
level before performing updates and examine the detailed logs.
Thanks,
---
JINMEI, Tatuya
Internet Systems Consortium, Inc.
More information about the bind10-dev
mailing list