Hi Abhijit and all,

The report for today:

LOW IMPACT:

- bug https://bugs.opendaylight.org/show_bug.cgi?id=1034 - OF statistics not stable. Statistics get delayed after OVS mininet restart. As I said not big issue but nice if this could be improved in the way is described in the bug.

- bug https://bugs.opendaylight.org/show_bug.cgi?id=947 - OSGi java errors/exception after starting mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

- bug https://bugs.opendaylight.org/show_bug.cgi?id=948 - OSGi java errors/exception after stopping mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

I really think you have addressed all issues found by integration group after datastore change so congratulations for that!

And now, looking forward, we need to focus on stability as AD-SAL OF plugin still shows more stable over multiple test runs. For this I will setup a call very soon (we are in the middle of a test case restructuring) to explain how to run the test suites locally, specially those test cases that are problematic.

BR/Luis 

On May 21, 2014, at 2:41 PM, Abhijit Kumbhare <abhijitkoss@...> wrote:

Good to know the following (from the integration report):

1. OpenFlow13 made some patches, more stable now, 100% test case passing in Robot (CASP3R, 15:05:24)
2. still some issue in stats for OpenFlow13 and some Flow issues that Chris and Luis working on (CASP3R, 15:05:50)

For the stats issue - are you referring to  https://bugs.opendaylight.org/show_bug.cgi?id=1034 - OF statistics not stable? I see a new comment on the bug by Luis:

This does not look like a big issue, it is just that statistics get delayed after restarting mininet OVS. This could be the result of the statistics optimization work. As an improvement I would ask to collect stats, not only when flows change but also when something changes in the inventory: add/remove switch or port.

Anil - what do you think?

On Tue, May 20, 2014 at 2:31 PM, Luis Gomez <ecelgp@...> wrote:

Hi all,

Big improvement from yesterday, this is the report from today:

FIXED:

- bug https://bugs.opendaylight.org/show_bug.cgi?id=995 - Controller does not learn node connectors from mininet OF10. Fixed as patch was merged https://git.opendaylight.org/gerrit/#/c/7127

- bug https://bugs.opendaylight.org/show_bug.cgi?id=1033 - Missing connectors first time we bring mininet OF13. Fixed with above patch as Kamal suspected.

HIGH IMPACT (test case fail):

- bug https://bugs.opendaylight.org/show_bug.cgi?id=1034 - OF statistics not stable. This happens with OVS mininet only (not seen with CPqD). I added TC to track this issue, I will be providing more logs for this as well. 

LOW IMPACT (only logs):

- bug https://bugs.opendaylight.org/show_bug.cgi?id=947 - OSGi java errors/exception after starting mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

- bug https://bugs.opendaylight.org/show_bug.cgi?id=948 - OSGi java errors/exception after stopping mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

BR/Luis

On May 19, 2014, at 6:24 PM, Luis Gomez <ecelgp@...> wrote:

Hi all,

These are the bugs detected today, also classified in high impact (produce test fail) or low impact (does not produce test fail):

HIGH IMPACT (test case fail):

- bug https://bugs.opendaylight.org/show_bug.cgi?id=995 - Controller does not learn node connectors from mininet OF10. Kamal/Moiz are working on this. They have have patch pending for review: https://git.opendaylight.org/gerrit/#/c/7127

- NEW bug https://bugs.opendaylight.org/show_bug.cgi?id=1033 - Missing connectors first time we bring mininet OF13. This can be fixed with above patch according to Kamal.

- NEW bug https://bugs.opendaylight.org/show_bug.cgi?id=1034 - OF statistics not stable. This explains the inconsistent OF13 test results we have seen today.

LOW IMPACT (only logs):

- bug https://bugs.opendaylight.org/show_bug.cgi?id=947 - OSGi java errors/exception after starting mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

- bug https://bugs.opendaylight.org/show_bug.cgi?id=948 - OSGi java errors/exception after stopping mininet: Michal working on this. Not fixed yet, we still see errors/exceptions

BR/Luis

On May 14, 2014, at 4:59 PM, Luis Gomez <ecelgp@...> wrote:

Hi all,

Looks like bug #967 is fixed as https://git.opendaylight.org/gerrit/#/c/6782/ was merged today. So now it is only bug #995 that is making our test RED, please prioritize this one.

BR/Luis



On May 14, 2014, at 3:22 PM, Abhijit Kumbhare <abhijitkoss@...> wrote:

We can discuss these in today's meeting.

- bug #947 to ofplugin-dev - OSGi java errors/exception after starting mininet: not fixed yet, work in progress

Michal R's fix (https://git.opendaylight.org/gerrit/#/c/6725/) was merged yesterday - please retry.

- bug #948 to ofplugin-dev - OSGi java errors/exception after stopping mininet: not fixed yet, work in progress

Michal R - this looks as if its still not been fixed. Can you check?

- bug #967 to ofplugin-dev - Flows do not show in operational area after deleting one flow using RESTCONF. This one is currently responsible for 6 consistent fails in the the test suite. There is a candidate patch in controller: https://git.opendaylight.org/gerrit/#/c/6782/ but it is still under review.

Ed has merged Tony's gerrit today. Please check now.

- NEW bug #995 to ofplugin-dev - Controller does not learn node connectors from mininet OF10. This new issue from today causes a lot of failures in the existing OF13 test suite.

Kamal/Moiz are working on this.

On Mon, May 12, 2014 at 7:55 PM, Luis Gomez <ecelgp@...> wrote:
Hi all,

The report of the day:

- bug #947 to ofplugin-dev - OSGi java errors/exception after starting mininet: not fixed yet, work in progress

- bug #948 to ofplugin-dev - OSGi java errors/exception after stopping mininet: not fixed yet, work in progress

- bug #967 to ofplugin-dev - Flows do not show in operational area after deleting one flow using RESTCONF. This one is currently responsible for 6 consistent fails in the the test suite. There is a candidate patch in controller: https://git.opendaylight.org/gerrit/#/c/6782/ but it is still under review.

- NEW bug #995 to ofplugin-dev - Controller does not learn node connectors from mininet OF10. This new issue from today causes a lot of failures in the existing OF13 test suite.

BR/Luis


On May 8, 2014, at 9:58 PM, Luis Gomez <ecelgp@...> wrote:

Hi devs, my report of the day:

- bug #947 to ofplugin-dev - OSGi java errors/exception after starting mininet: not fixed yet, work in progress

- bug #948 to ofplugin-dev - OSGi java errors/exception after stopping mininet: not fixed yet, work in progress

- bug #933 to ofplugin-dev - AD-SAL topology in GUI or REST does not get cleared after stopping mininet: Fixed in controller through: https://git.opendaylight.org/gerrit/#/c/6785

- bug #967 to ofplugin-dev - Flows do not show in operational area after deleting one flow using RESTCONF. This one is currently responsible for 6 consistent fails in the the test suite. There is a candidate patch in controller: https://git.opendaylight.org/gerrit/#/c/6782/ but it is still under review.

BR/Luis



On May 6, 2014, at 8:50 PM, Luis Gomez <ecelgp@...> wrote:

And the update for today:

- bug #947 to ofplugin-dev - OSGi java errors/exception after starting mininet: it is not fixed yet, i get even more errors in the OSGi console, it is reported in the bug. Madhusudhan opened similar bug #962 to controller-dev

- bug #948 to ofplugin-dev - OSGi java errors/exception after stopping mininet: it is not fixed yet, i get different errors in the OSGi console, it is reported in the bug. Madhusudhan opened similar bug #963 and #966 to controller-dev

- bug #933 to ofplugin-dev - AD-SAL topology in GUI or REST does not get cleared after stopping mininet: not fixed yet

- bug #967 to ofplugin-dev - Flows do not show in operational area after deleting one flow using RESTCONF, this is the issue I commented yesterday

BR/Luis



On May 5, 2014, at 5:20 PM, Luis Gomez <ecelgp@...> wrote:

Hi all,

An update on the below issues:

- bug #937 to controller-dev to fix OSGi java exception after new datastore: seems to be fixed now

- bug #947 to ofplugin-dev to fix OSGi java exception after mininet starts: waiting for the patch https://git.opendaylight.org/gerrit/#/c/6725/

- bug #948 to ofplugin-dev to fix OSGi java exception after mininet stops: fixed the existing exception but we get a new one reported in the same bug

On top of this we realized AD-SAL topology in GUI or REST does not get cleared after stopping mininet. This has been reported in bug #933.


BR/Luis



On May 2, 2014, at 5:34 PM, Luis Gomez <ecelgp@...> wrote:

Hi ofplugin people,

The other day call I was showing a graph to compare AD-SAL plugin vs MD-SAL plugin behavior on the same test suite. Now I realize that even before we go and analyze the TCs that are not stable in of13, we need to fix the existing OSGi errors produced during the test run:

https://jenkins.opendaylight.org/integration/view/Deploy%20Jobs/job/integration-deploy-controller-latest-base-of13-edition/ws/opendaylight/runlog.txt

For the  simple reason that some of these errors could be behind the stability issues in the test suite.

So far we have opened the following bugs:

- bug #937 to controller-dev to fix OSGi java exception after new datastore. This is really annoying bug as this recurrent exception is getting mixed with ofplugin exceptions

- bug #947 to ofplugin-dev to fix OSGi java exception after mininet starts

- bug #948 to ofplugin-dev to fix OSGi java exception after mininet stops

There will be more bugs coming (if you see the OSGi console printout from Jenkins) but lets start with these 3.

Thanks/Luis

_______________________________________________
openflowplugin-dev mailing list
openflowplugin-dev@...
https://lists.opendaylight.org/mailman/listinfo/openflowplugin-dev

_______________________________________________
controller-dev mailing list
controller-dev@...
https://lists.opendaylight.org/mailman/listinfo/controller-dev