Hello Brady

I looked also at the failure at RC2 csit build 116 [1] and got an idea
of what is going on (applies to the carbon SR2 too). It looks like the
create RSP rpc is being invoked before the SF listener has been
trigerred to add the SF type information dynamically. Thus the RSP
scheduler does not find SF by type as that info has not been added yet
to the database.

I noticed also during a test, there is still listeners being trigered
corresponding to actions of the previous a test. This whole thing
creates some unexpected database states for which SFC is not prepared
to handle well thus we have a bunch of NPEs.

So there is some hardening bit to do in both areas.

BR
Jaime.




I thing the 

Tehre is interleaved activity on 2 nodes of the cluster a 
On jue, 2017-08-17 at 20:07 +0000, Brady Johnson wrote:
> Jaime,
> 
> Thanks for the analysis!!
> 
> Regards,
> 
> Brady
> 
> 
> On Thu, Aug 17, 2017 at 12:34 PM Jaime Caamaño Ruiz <
> [email protected]> wrote:
> 
> > 
> > Hello
> > 
> > I looked into this and while I cannot pinpoint a precise cause, I
> > think
> > it was due to a sporadic database error. Some observations:
> > 
> > - Test fails because it cannot create an RSP. RSP cannot be created
> >   because cant read SF types from DB for the SF that wants to use.
> > SF
> >   types are created as SF are configured, which happenned
> > previously
> >   with no issue.
> > - There is interleaved application activity in cluster node 1 and
> >   cluster node 3 at the time of failure, probably indicating some
> > other
> >   problem going on at the time.
> > - Tested the same procedure locally on the same release and worked.
> > - Checked RC2 changes and none looks like could be related with
> > this
> >   issue.
> > 
> > Brady, I think we can say this is OKAY.
> > 
> > BR
> > Jaime.
> > 
> > On mié, 2017-08-16 at 23:02 +0000, Brady Johnson wrote:
> > > 
> > > An,
> > > 
> > > Thanks for the info, Im looking into it now.
> > > 
> > > Regards,
> > > 
> > > Brady
> > > 
> > > 
> > > On Wed, Aug 16, 2017 at 12:44 PM An Ho <[email protected]> wrote:
> > > 
> > > > 
> > > > 
> > > > Hi SFC Team,
> > > > 
> > > > 
> > > > 
> > > > There are Distribution Test failures [1] in the Carbon SR2
> > > > Build
> > > > 20170815
> > > > [2] related to your project.  Please take a moment to mark
> > > > these
> > > > issues in
> > > > the spreadsheet [1] as "OKAY" to release or "BLOCKING" and
> > > > requires
> > > > a
> > > > respin of Carbon SR2.
> > > > 
> > > > 
> > > > 
> > > > Best Regards,
> > > > 
> > > > An Ho
> > > > 
> > > > 
> > > > 
> > > > [1]
> > > > https://docs.google.com/spreadsheets/d/1VcB12FBiFV4GAEHZSspHBNx
> > > > KI_9
> > > > XugJp-6Qbbw20Omk/edit#gid=1226505712
> > > > 
> > > > [2]
> > > > https://wiki.opendaylight.org/view/Simultaneous_Release:Carbon_
> > > > Rele
> > > > ase_Plan#SR2_Download
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > _______________________________________________
> > > sfc-dev mailing list
> > > [email protected]
> > > https://lists.opendaylight.org/mailman/listinfo/sfc-dev
_______________________________________________
sfc-dev mailing list
[email protected]
https://lists.opendaylight.org/mailman/listinfo/sfc-dev

Reply via email to