On 7/19/22 11:14, Stokes, Ian wrote:
>> On 7/15/22 20:33, Ilya Maximets wrote:
>>> On 7/15/22 20:24, Stokes, Ian wrote:
>>>>> Looks like new MTU tests are unstable.
>>>>> Ian, Michael, could you, please, check?
>>>>>
>>>>> Best regards, Ilya Maximets.
>>>>
>>>> Hey Ilya,
>>>>
>>>> Let me take a look at this, it passed our internal CI so not sure why it's
>> failed externally, worst case I can revert this as it was really just a nice 
>> to
>> have, if it's holding up the branching would you prefer this?
>>>
>>> No, it's not holding up branching.  Just wanted to highlight the
>>> issue.  Would be great to fix it in a near future, but it's not
>>> critical for right now.  Thanks.
>>
>> I think, the tests are lacking the 'Clean up the testpmd now' part
>> at least.  If testpmd is still running while we're tearing down OVS,
>> there is a fair chance that some packets will be received while the
>> OpenFlow port is already removed.  That migth explain some of the
>> failures.  Not sure if that what happened in this particular case
>> thouhg.
>>
>> Also, looksing at other failure reports I see some weird things like:
>>
>>   Unreasonably long 1140ms poll interval (408ms user, 263ms system)
>>
>> 408 + 263 != 1140.  The main thread got preempted for half a second
>> in this case.   Are you running something else on the same cores on
>> that system?  Or maybe some processes are not getting stopped properly
>> after the test?
> 
> Thanks Ilya, I'll look at the unit tests today (apologies for the delay,  few 
> folks were out of office yesterday).
> With regards the system, Michael will be back in office tomorrow and I can 
> ask him to check the system then to see if there are any processes acting 
> incorrectly?

Sure.  Thanks for the update.

Best regards, Ilya Maximets.
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to