CSIT Sporadic failures - ElanService Suite - ssh instance connectivity fails after ovs restart

Description

at a high level, the ovs instances on both compute nodes are restarted. The
elan service flows appear to be validated as correct on both computes after
the restart. Then an ssh to an openstack instance from the control node's
dhcp namespace fails:

https://logs.opendaylight.org/releng/jenkins092/netvirt-csit-1node-openstack-newton-upstream-learn-carbon/233/archives/log.html.gz#s1-s5-s1-t2-k13-k5

Environment

Operating System: All
Platform: All

Attachments

Linked issues

is duplicated by

NETVIRT-910

Flow is not deleted in Controller Node after Virtual NW deleted.

Activity

Vivekanandan Narasimhan June 13, 2017 at 11:56 AM

Ok Peri,

but like I noted in that email [0] that it hadn't happened in a while, I still don't think it's happened since then.

anyway, it sounded like you were waiting on me for something so I wanted to make sure I wasn't blocking anything.

JamO

On 06/07/2017 10:57 PM, Periyasamy Palanisamy wrote:
> Yes JamO, I was referring to instance restart test case for OF port number not changing after VM reboot. Due to this, instance restart test case will get marked as failure.
> So either we have to change VM reboot procedure as hard reboot or need to figure out why soft reboot is not working.
>
> Thanks,
> Periyasamy
>
> ----Original Message----
> From: Jamo Luhrsen jluhrsen@gmail.com
> Sent: Thursday, June 08, 2017 2:31 AM
> To: N Vivekanandan <n.vivekanandan@ericsson.com>; Periyasamy
> Palanisamy <periyasamy.palanisamy@ericsson.com>
> Subject: Re: Bug - https://bugs.opendaylight.org/show_bug.cgi?id=8302
>
> Hi Peri,
>
> I think we figured out the OF number not changing after the reboot.
> At least I analyzed it before and figured it was because sometimes
> the "soft reboot" doesn't work and then only after 2m (or so) nova
> finally does a "hard reboot". see email thread here [0]
>
> But, is that what you are referring to below? Todays failure report had the "ovs restart" bug, not the "instance restart" bug.
>
> Thanks,
> JamO
>
> [0]
> https://lists.opendaylight.org/pipermail/netvirt-dev/2017-May/004543.h
> tml
>
>
> On 06/07/2017 11:39 AM, N Vivekanandan wrote:
>> Hi Jamo,
>>
>>
>>
>> From your digest mail today, it looks like 8302 is still happening.
>>
>>
>>
>> Related to that Peri has a question below, can you please help with your inputs?
>>
>>
>>
>> Vivek
>>
>>
>>
>> From: Periyasamy Palanisamy
>> Sent: Wednesday, June 07, 2017 10:02 AM
>> To: N Vivekanandan <n.vivekanandan@ericsson.com>
>> Subject: RE: Bug -
>> https://bugs.opendaylight.org/show_bug.cgi?id=8302
>>
>>
>>
>> This issue was analyzed already and merged two patches for table 17 stale entries issue (one patch is merged yesterday).
>>
>> Let us wait for fresh CSIT report to see whether this issue happens again or not.
>>
>>
>>
>> JamO has to come back on why OF port number is not changed after nova reboot.
>>
>>
>>
>> Thanks,
>>
>> Periyasamy
>>
>>
>>
>> From: N Vivekanandan
>> Sent: Wednesday, June 07, 2017 12:02 AM
>> To: Periyasamy Palanisamy <periyasamy.palanisamy@ericsson.com
>> <periyasamy.palanisamy@ericsson.com>>
>> Subject: Bug - https://bugs.opendaylight.org/show_bug.cgi?id=8302
>>
>>
>>
>> Hi Peri,
>>
>>
>>
>> Seems to be related to ELAN as DHCP IP assignment is failing (reported by Jamo today).
>>
>>
>>
>> Can we have someone take a look ?
>>
>>
>>
>> –
>>
>> Thanks,
>>
>>
>>
>> Vivek
>>
>>
>>