>With a simple loop sending the same query to the backend via nc, I
>can do a little over 16 queries/second from my laptop. This rises to
>a little over 18/s when run from the backend and sending the queries
>to localhost.

OK, a bit of tuning has upped that to 23 & 36 respectively - more if 
I use an IP address rather than FQDN. Furthermore, if I run the same 
stress test from 5 clients at one, the backend seems to saturate (no 
idle time, 1 or 2% WIO) at about 50 queries/s total.

Tuning changes I've made are :
key_buffer                 16M -> 128M
thread_cache_size            8 -> 32
table_cache                 64 -> 128

innodb_flush_log_at_trx_commit -> 2
join_buffer_size               -> 192K
tmp_table_size                 -> 64M

Between them, these seem to have vastly improved the performance of 
the DB - though I need to do some pruning back as mysqltuner now 
imforms me that the max usable memory is more than the physical 
memory installed in the machine.
mysqltuner  still reports "Temporary tables created on disk: 21%" 
which seems a bit high.

I'm still seeing the same thing happening - the mailing list server 
is sending out two emails, then blocking for a while (several 
minutes) after the mail handler reports the same errors (table lookup 
problem).

On the upside, I'm learning about tuning params for MySQL !

I'm starting to wonder if I shouldn't load MySQL on each mail handler 
and replicate the Postfix Admin DB to each mail handler - I know it's 
only masking the issue, but it should stop the errors :-/

-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.
_______________________________________________
Users mailing list
[email protected]
http://lists.policyd.org/mailman/listinfo/users

Reply via email to