Tq time fluctuates wildly


#1

Experiencing high fluctuation in Tq times.

1317/0/1/3/1321
130/0/0/92/230
55/0/0/240/295
106/0/0/3/109
106/0/0/2/108
6260/0/0/68/6328
974/0/1/4/979
1592/0/1/93/3501
75/0/0/1525/1600
123/0/1/2/126
124/0/0/3/127
10332/0/0/241/10574
758/0/0/3/761
40/0/0/1535/1576
35/0/0/120/155

This is running HAProxy 1.5 with native SSL and also seeing the same using stunnel.

Not using option http-server-close so this is not keep-alive and time between requests.

Running with option forceclose to get accurate counts.

What can cause such variability, not shown but some go as high at 30, 40, 50, 60+ seconds.


#2

Hi,
Have you resolved this issue? We are running HAProxy 1.5.1 behind AWS ELB and seeing the same thing. Tq fluctuates wildly for no reason, it’s not related to heavy traffic, or any particular backend boxes, up to 50% of requests wasting 100ms in Tq times.

Did you upgrade to later version to address the issue? after all, 1.5.1 was 3 years old and may not play nice with AWS ELB.