ssh connection getting dropped frequently

[Jagatjot Singh]

Openstack version - pike Installation process - multinode kolla

We are frequently facing ssh connection loss on our VMs which are running in our production environment. RAM memory usage on our controller nodes becomes low because of which memory automatically shifts to swap. Moreover we checked the CPU utilization of neutron-l3-agents by running docker stats command. Neutron-l3-agent cpu utilization gives lots of spikes varying from 25% to 250%. We have also verified all the services of openstack and all the services are running fine without showing any errors in the logs. However ssh connections on all the VMs gets dropped after 5-10 seconds without giving any error. Could you confirm us the exact reason why we are facing this issue ?

[Jagatjot Singh (jagat.singh)]

I have also attached the screenshot highlighting the CPU utilization of Neutron-L3-agent.

[Brian Haley (brian-haley)]

Unfortunately the image you attached doesn't give much info on what is going on.

Are there failure messages in the l3-agent or neutron-server logs?

Pike is quite old, can you reproduce this on a more recent version?

[Jagatjot Singh (jagat.singh)]

The attached images shows the CPU utilization of l3-agent which varies from 25% to 250%.

No, there are no failures messages in the l3-agent or neutron-server logs.

Our production environment is using Pike version and we are facing this issue on our production environment.

[Brian Haley (brian-haley)]

Unfortunately the image you provided doesn't help narrow-down the issue, more info from the logs would be necessary.

You could try enabling debug (debug=True in l3_agent.ini) and see if that gives more data.

Since this is a production system it might be best to contact your vendor, especially since Pike is older, they might be able to help with the issue as well.

[Launchpad Janitor (janitor)]

[Expired for neutron because there has been no activity for 60 days.]