Roi Dayan <[email protected]> writes:

> On 8/30/24 16:00, Aaron Conole wrote:
>> Eelco Chaudron <[email protected]> writes:
>> 
>>> It is observed in some environments that there are much more ukeys than
>>> actual DP flows. For example:
>>>
>>> $ ovs-appctl upcall/show
>>> system@ovs-system:
>>> flows : (current 7) (avg 6) (max 117) (limit 2125)
>>> offloaded flows : 525
>>> dump duration : 1063ms
>>> ufid enabled : true
>>>
>>> 23: (keys 3612)
>>> 24: (keys 3625)
>>> 25: (keys 3485)
>>>
>>> The revalidator threads are busy revalidating the stale ukeys leading to
>>> high CPU and long dump duration.
>>>
>>> This patch tracks the number of consecutive missed dumps. If four dumps
>>> are missed in a row, it is assumed that the datapath flow no longer
>>> exists, and the ukey can be deleted.
>>>
>>> Reported-by: Roi Dayan <[email protected]>
>>> Co-authored-by: Han Zhou <[email protected]>
>>> Co-authored-by: Roi Dayan <[email protected]>
>>> Signed-off-by: Han Zhou <[email protected]>
>>> Signed-off-by: Roi Dayan <[email protected]>
>>> Signed-off-by: Eelco Chaudron <[email protected]>
>>> ---
>> Thanks all, applied.
>> 
>
> Hi,
>
> Can we get this merged into 3.x branches maybe ?

Yes, I think it makes sense there.  While there isn't a specific 'fixes'
for this, it does seem more like a fix rather than a feature.

> From 3.0 to 3.3 the python script and flow_del_reason enum
> doesn't exists. With removing the diff and removing
> "del_reason = FDR_FLOW_MISSING_DP;" should apply cleanly.

It will fail because it depends on some other support (for example
strip_dp_hash in the tests).  I'm working on a backport for it, and will
post.

> This patch applies cleanly without modification on branch 3.4.
>
> Tell me if needed to send a back-ported patch that can be applied
> to branches 3.0-3.3.

I will be posting it for each branch when it is ready - I think I should
be happy with it all by COB tomorrow.

> Thanks,
> Roi
>
>> _______________________________________________
>> dev mailing list
>> [email protected]
>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev
>
> _______________________________________________
> dev mailing list
> [email protected]
> https://mail.openvswitch.org/mailman/listinfo/ovs-dev

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to