Lastchk showing high values



Facing some slow responses on one of the application.

The Lastchk value on status page is showing upto 2000ms at times and in similar setup another haproxy the Lastchk is ‘0’ (zero).

The haproxy configuration is exact same (except servers and VLAN).

It may be a network issue or very high load.
How haproxy check is performed ? (the ping values from harpoxy to server are fine).



The amount it takes for the health check to complete. What that health check does, depends on how you configured it.

For example, if you access a specific URL on the server for the health check and to return the response the servers make some internal checks (like connecting to a database), then all those steps add to that value (as the health check waits for a response).

Make the same exact call externally, and it should yield the same results.


To add a few relevant details of haproxy.cfg

HA-Proxy version 1.5.18 2016/05/10

mode http

server SERVER_101 XX.XX.XX.XX.:80 check cookie server_101 maxconn 3000

There is no specific health check.

How to call the default health check externally?


By externally I mean not with haproxy, but with an external programm like curl for example.

If you don’t have any httpchecks configured at all, haproxy will only establish the TCP connection (3-way handshake). If this already takes a long time, then there is probably something going on at TCP level. How do the Tc counters in your logs look like?


Tc is high and goes higher with time and load.

How exactly to do a 3-way handshake. Telnet ??
I am not very familiar with network commands.

If I can simulate the handshake I might be able to capture something in tcpdump to find the problem.



You can try a tcp traceroute:

sudo traceroute -T -p 80 <destination IP>


Thanks !
let me try.