Hi,
I'm using HAProxy in production on EC2 virtual machines at AWS. These machines
are behind an ELB (Elastic Load Balancer). Working our on performances, we
couldn't but notice that our typical values for Tq are very high:
[14:53][prod] root@f04:~# awk '{print $10;}' /var/log/haproxy_info.log | cut
-d'/' -f1 | wc -l
341444
[15:03][prod] root@f04:~# awk '{print $10;}' /var/log/haproxy_info.log | cut
-d'/' -f1 | numaverage
430.665283544852
First number is the number of processed log lines. 430ms is the average for Tq.
A bit high isn't it ?
Do you have any idea of how we can work on the issue a bit further, knowing
that the ELB is a black box on which we have little to no control ?
Cheers,
--
Pierre <[email protected]>
Server Shepherd at http://www.fotopedia.com/
