[L2][scale issue] local connection to ovs-vswitchd was drop or timeout
Bug #1813705 reported by
LIU Yulong
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
When subnets or security group ports quantity reach 2000+, the ovs-agent connection to ovs-vswitchd may get lost, drop or timeout during restart.
This is a subproblem of bug #1813703, for more information, please see the summary:
https:/
tags: | added: neutron-proactive-backport-potential |
tags: | removed: neutron-proactive-backport-potential |
Changed in neutron: | |
status: | New → Fix Released |
To post a comment you must log in.
Are we running out of some OS resources and is that the reason for the ovs-agent connection to get dropped to the ovs-vswitchd.
If that is the case, is it possible for us to throttle the number of ports for initial sync when the agent comes up.
We did similar throttling mechanism for the number of routers that are handled by the l3-agents, may be a similar approach to throttle the amount of ports on a particular node may be solution.
Again if the logs are not communicating to us anything about the issue, we should probably also update the logs to communicate the exact problem.