WiFi latency issues with TPLink C2600 routers in one direction only

Hi,

We use a C2600 as our main router and a second C2600 as an access point. Both use 21.02.3. AP has a static IP outside the DHCP range of the main router and is wired through gigabit to it. Both have WiFi active but on different bands.

We used to have no issues... can't say when exactly they started, or if it happened after a version upgrade. Up 'till now I didn't have any metrics to show something was wrong. I just came with this up today.

Main router is 192.168.12.1. Everything here is on the 149 5GHz band. We only use 2.4GHz for a wireless printer.

Pinging the AP from a linux workstation (192.168.12.127) directly connected to the AP (192.168.12.80) shows expected latency, average 3ms, rarely if ever above 5ms. Sample:

64 bytes from AP (192.168.12.80): icmp_seq=586 ttl=64 time=3.32 ms
64 bytes from AP (192.168.12.80): icmp_seq=587 ttl=64 time=1.34 ms
64 bytes from AP (192.168.12.80): icmp_seq=588 ttl=64 time=2.96 ms
64 bytes from AP (192.168.12.80): icmp_seq=589 ttl=64 time=3.71 ms
64 bytes from AP (192.168.12.80): icmp_seq=590 ttl=64 time=3.29 ms
64 bytes from AP (192.168.12.80): icmp_seq=591 ttl=64 time=4.06 ms
64 bytes from AP (192.168.12.80): icmp_seq=592 ttl=64 time=2.98 ms
64 bytes from AP (192.168.12.80): icmp_seq=593 ttl=64 time=4.02 ms

At the same time, pinging the other way, from the AP to the linux workstation, is a mess:

64 bytes from 192.168.12.127: seq=589 ttl=64 time=31.798 ms
64 bytes from 192.168.12.127: seq=590 ttl=64 time=54.296 ms
64 bytes from 192.168.12.127: seq=591 ttl=64 time=283.435 ms
64 bytes from 192.168.12.127: seq=592 ttl=64 time=102.359 ms
64 bytes from 192.168.12.127: seq=593 ttl=64 time=126.114 ms
64 bytes from 192.168.12.127: seq=594 ttl=64 time=150.824 ms
64 bytes from 192.168.12.127: seq=595 ttl=64 time=174.256 ms
64 bytes from 192.168.12.127: seq=596 ttl=64 time=197.727 ms
64 bytes from 192.168.12.127: seq=597 ttl=64 time=2.562 ms
64 bytes from 192.168.12.127: seq=598 ttl=64 time=40.151 ms
64 bytes from 192.168.12.127: seq=599 ttl=64 time=63.417 ms
64 bytes from 192.168.12.127: seq=600 ttl=64 time=87.443 ms
64 bytes from 192.168.12.127: seq=601 ttl=64 time=110.941 ms
64 bytes from 192.168.12.127: seq=602 ttl=64 time=133.809 ms
64 bytes from 192.168.12.127: seq=603 ttl=64 time=157.290 ms
64 bytes from 192.168.12.127: seq=604 ttl=64 time=181.250 ms
64 bytes from 192.168.12.127: seq=605 ttl=64 time=159.853 ms
64 bytes from 192.168.12.127: seq=606 ttl=64 time=24.010 ms
64 bytes from 192.168.12.127: seq=607 ttl=64 time=1.719 ms
64 bytes from 192.168.12.127: seq=608 ttl=64 time=70.625 ms
64 bytes from 192.168.12.127: seq=609 ttl=64 time=504.299 ms
64 bytes from 192.168.12.127: seq=610 ttl=64 time=41.673 ms
64 bytes from 192.168.12.127: seq=611 ttl=64 time=346.664 ms
64 bytes from 192.168.12.127: seq=612 ttl=64 time=1.656 ms

Happens with other devices too.

Ping times between two WiFi clients are usually equally bad or worse. This is a sample from linux workstation to another linux box, both WiFi clients connected to AP.

64 bytes from M90 (192.168.12.166): icmp_seq=507 ttl=64 time=348 ms
64 bytes from M90 (192.168.12.166): icmp_seq=508 ttl=64 time=370 ms
64 bytes from M90 (192.168.12.166): icmp_seq=509 ttl=64 time=4.72 ms
64 bytes from M90 (192.168.12.166): icmp_seq=510 ttl=64 time=416 ms
64 bytes from M90 (192.168.12.166): icmp_seq=511 ttl=64 time=31.6 ms
64 bytes from M90 (192.168.12.166): icmp_seq=512 ttl=64 time=53.2 ms
64 bytes from M90 (192.168.12.166): icmp_seq=513 ttl=64 time=75.4 ms
64 bytes from M90 (192.168.12.166): icmp_seq=514 ttl=64 time=99.0 ms
64 bytes from M90 (192.168.12.166): icmp_seq=515 ttl=64 time=323 ms
64 bytes from M90 (192.168.12.166): icmp_seq=516 ttl=64 time=143 ms

There is one curious exception though. We have another OpenWRT device that we use as a wireless camera. Ping times between that and the linux workstation are stellar.

64 bytes from Cam1 (192.168.12.81): icmp_seq=511 ttl=64 time=6.39 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=512 ttl=64 time=9.03 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=513 ttl=64 time=3.01 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=514 ttl=64 time=2.87 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=515 ttl=64 time=3.93 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=516 ttl=64 time=6.57 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=517 ttl=64 time=4.97 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=518 ttl=64 time=7.09 ms
64 bytes from Cam1 (192.168.12.81): icmp_seq=519 ttl=64 time=2.96 ms

All computers are having it easy with high idle CPU and network was not busy at all when I ran these pings.

Any suggestions what I could look at next to make this better?

Thanks in advance.

Ugh. Figured it out. Not an OpenWRT problem.

The linux server and workstations have WiFi power management on. An update must have brought that in.

This explains it: Linux WiFi DPM Latency.

Leaving it here for when I trip on this again.

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.