Dear Haproxy Users,
I would like to ask some help with a strange behaviour, which I read that maybe cause the haproxy’s timeout values. First I appologise for bad english and also for I’m not heavy user of haproxy.
So the problem: I use haproxy in tcp mode to proxy ms sql queries. The config a simple plain like:
defaults
log global
mode tcp
retries 3
maxconn 256
timeout connect 5s
timeout client 15m (was 30s, but I raised for testing)
timeout server 15m (was 30s, but I raised for testing)
default-server check inter 5s
listen sql
bind 10.10.10.1:1433
mode tcp
option tcplog
log global
maxconn 256
server sqlbe 192.168.5.5:1433
The mssql administrator tells me that a lot of (40-50) “empty” begin/commit transaction appears in the log file at the same time (in 1 sec). The begin/commit logs not in pair, for example 3 begins, 2 commits, 1 begin, 3 commits, 2 begins …
The timeouts were 30secs, but I read that too small timeouts can be cause this kind of behaviour.
Because I don’t know deeply what the problem is, I would like to ask is there anybody who experienced such a thing? Is there any idea what could be the problem?
Thank you very much
1 post - 1 participant