vppapi event takes a long time

[glazychev-art]

Description

Looking through the logs, it was noticed that the event from vpp about upping memif interface can take a long time.
See: https://github.com/networkservicemesh/sdk-vpp/blob/main/pkg/networkservice/up/common.go#L102-L118

For example, we were supposed to get an event here:

...
{"log":"Nov 24 21:55:25.958\u001b[37m [DEBU] [id:15381945-9fe7-4afa-8b2b-f0e6347af950] [name:forwarder-vpp-4rq6w] [details.Flags:IF_STATUS_API_FLAG_ADMIN_UP] [duration:257.36µs] [swIfIndex:4] [type:NetworkService] [vppapi:SwInterfaceDump] \u001b[0m(48.2)                                                  completed span=7e3a66648206cc0f:28c2658fae6b3450:705fd3ed70c65f2b:1\n","stream":"stderr","time":"2021-11-24T21:55:25.959022416Z"}
{"log":"Nov 24 21:55:33.330\u001b[37m [TRAC] [id:15381945-9fe7-4afa-8b2b-f0e6347af950] [name:forwarder-vpp-4rq6w] [type:NetworkService] \u001b[0m(48.3)                                                  close={...}
...

Later we get an event, but it takes a long time (see duration):

...
{"log":"Nov 24 21:56:03.055\u001b[37m [DEBU] [id:ecbdebf2-c694-41e0-9e02-51ebaacde72b] [name:forwarder-vpp-4rq6w] [duration:9.059329171s] [msg.Flags:IF_STATUS_API_FLAG_LINK_UP] [swIfIndex:4] [type:NetworkService] [vppapi:SwInterfaceEvent] \u001b[0m(48.3) 
...

Logs

forwarder21.zip

Related issue

networkservicemesh/deployments-k8s#2381

[glazychev-art]

Hi @jkossak

We looked at the logs from nse-composition case when ping did not work for you (forwarder log is attached in description)
Could you please clarify how often this problem is? Do you have logs from other runs?

[jkossak]

Hi @glazychev-art, it rather happened on my system when I was testing certain commit. When I turned back to try previous commmits the ping was working again. I returned to the same one and ping was not working again.

Now I'm on commit b8988399c800a966cf8148d0e42762e54266f6c9 and I do not not see issues with ping test.

The commit that I used when ping did not work was f04abcdc94c1a96652886c1d84f6e29601431ea4. From that commit I removed all cpu and mem limits and requests for exclude-prefixes to avoid restarts of exclude-prefixes container.

[denis-tingaikin]

Seems like the issue outdated and was fixed by f04abcdc94c1a96652886c1d84f6e29601431ea4.

@jkossak Be free to reopen this if you see any issues.