High CPU-load with SFOS 17.1.2 MR-2 and HA

This issue occurs only if the RED feature is enabled and at least one RED interface is connected.

The more RED interfaces are used, the heavier the issue occurs. Also happens on non-HA setups.

It will be fixed in v17.1.MR3.

This issue occurs only if the RED feature is enabled and at least one RED interface is connected.

The more RED interfaces are used, the heavier the issue occurs. Also happens on non-HA setups.

It will be fixed in v17.1.MR3.

Which leads to the next question when is MR3 planned for release?

Ian

Hey Ian,

Unfortunately, no release date news available yet. Please stay tuned for the latest.

To update this thread for our community:

This KB article has been published regarding this issue.

Regards,

SFOS 17.1.3 MR3 Released

Not sure this bug is fixed in MR-3 ...  High CPU fix included is this one: NC-36927 [RED] High CPU caused by gethainfo.

Paul Jr

I tried the update on a XG125 HA-Cluster today.

But the heavy load is still there, like in MR2.

Now it's not the HA-process, there are a lot of worker processes.

Current load with no traffic and about 25 RED-tunnel ~4.10

And now?

EDIT 2h after the update:

top - 14:23:20 up 1:47, 1 user, load average: 2.99, 2.87, 3.11
Tasks: 412 total, 1 running, 410 sleeping, 0 stopped, 1 zombie
Cpu(s): 0.8%us, 0.2%sy, 0.0%ni, 98.8%id, 0.0%wa, 0.0%hi, 0.2%si, 0.0%st
Mem: 4010540k total, 3709820k used, 300720k free, 190656k buffers
Swap: 2029868k total, 0k used, 2029868k free, 1057480k cached

Looks like idle is okay, load is still high?

EDIT the next morning:

I took a look at the LOAD AVERAGE from the past 4 weeks, and the crazy thing is load was minimal higher with SFOS 17.1.1 MR-1, now it's about ~2,9, but the system has a slow and not well performing feeling - I mean only the WebAdmin - services are performing well.