Re: [controller-dev] OFP cluster test with "tell-based"

[Robert Varga]

On 26/02/2019 21:52, Luis Gomez wrote:
> FYI tell-based test with latest master:
> 
> https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-3node-gate-clustering-only-sodium/9/
> 
> I still see topology link information is not cleared after node/link down in 
> some random tests.

There does not seem to be anything out of the ordinary from CDS
perspective, but there are some OLFE/validation errors.

Can someone from the OFP team take a look, please?

Thanks,
Robert



signature.asc
Description: OpenPGP digital signature
___
controller-dev mailing list
controller-dev@lists.opendaylight.org
https://lists.opendaylight.org/mailman/listinfo/controller-dev

Re: [controller-dev] OFP cluster test with "tell-based"

[Luis Gomez]

FYI tell-based test with latest master:

https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-3node-gate-clustering-only-sodium/9/

I still see topology link information is not cleared after node/link down in 
some random tests.

BR/Luis

> On Feb 20, 2019, at 2:35 AM, Robert Varga  wrote:
> 
> On 19/02/2019 19:50, Luis Gomez wrote:
>> 
>> 
>>> On Feb 19, 2019, at 6:16 AM, Robert Varga >> > wrote:
>>> 
>>> On 19/02/2019 02:11, Luis Gomez wrote:
 
 
> On Feb 13, 2019, at 2:22 AM, Robert Varga  > wrote:
> 
> On 12/02/2019 19:44, Luis Gomez wrote:
>> Hi everybody,
>> 
>> FYI I have just tried OFP cluster test with "tell-based" protocol:
>> 
>> https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/openflowplugin-csit-3node-clustering-only-neon/180/robot-plugin/log.html.gz
>> 
>> My observations:
>> 
>> 1) node/port down events do not clear links in topology, this is
>> why all topology check test fail.
> 
> I think this is related to the transactions not commit in 5 seconds,
> hence masters are not created.
 
 Any workaround for this?
>>> 
>>> Not sure... if we have messed up accounding (below), we may end up
>>> reporting things out of whack.
> 
> Alright, we can ditch the builder debugs, as everything there works as
> it is supposed to.
> 
> The reason for non-removal of the links is captured in
> https://logs.opendaylight.org/sandbox/vex-yul-odl-jenkins-2/openflowplugin-csit-3node-clustering-only-sodium/1/odl_3/odl3_karaf.log.gz,
> I think, and it is a VerifyException.
> 
> Filed to https://jira.opendaylight.org/browse/CONTROLLER-1885.
> 
> Regards,
> Robert
> 

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

[controller-dev] 2019.02.26 Kernel Projects weekly meeting minutes

[Robert Varga]

Hello,

the minutes for this meeting are available at:

> [26/02/2019 18:58:58]  Meeting ended Tue Feb 26 17:58:58 2019 
> UTC.  Information about MeetBot at http://ci.openstack.org/meetbot.html . (v 
> 0.1.4)
> [26/02/2019 18:58:58]  Minutes:
> http://meetings.opendaylight.org/opendaylight-meeting/2019/kernel_projects_call/opendaylight-meeting-kernel_projects_call.2019-02-26-17.06.html
> [26/02/2019 18:58:58]  Minutes (text): 
> http://meetings.opendaylight.org/opendaylight-meeting/2019/kernel_projects_call/opendaylight-meeting-kernel_projects_call.2019-02-26-17.06.txt
> [26/02/2019 18:58:58]  Log:
> http://meetings.opendaylight.org/opendaylight-meeting/2019/kernel_projects_call/opendaylight-meeting-kernel_projects_call.2019-02-26-17.06.log.html

and also as an attachment to this message.

Regards,
Robert

Title: #opendaylight-meeting: Kernel Projects call





#opendaylight-meeting: Kernel Projects call

Meeting started by rovarga at 17:06:03 UTC
(full logs).





Meeting summary

odlparent-5.0.0 (rovarga, 17:06:08)

  https://jira.opendaylight.org/projects/ODLPARENT/versions/10825
is the tracker (rovarga,
17:06:16)
  most of the targeted issues are in
review (rovarga,
17:06:27)
  there are couple of patches we want in 4.0.10,
will wait for them to land before branching (rovarga,
17:06:44)
  not sure about
https://jira.opendaylight.org/browse/ODLPARENT-176 (rovarga,
17:08:26)
  the bundle name is useful in general
(rovarga,
17:08:37)
  and this can always be tuned in a particular
deployment (rovarga,
17:08:49)
  most notably, high-log-deployments will
probably end up using logstash or something of that kind to capture
the logs (and sift through them) (rovarga,
17:09:17)


karaf-4.2.3 (rovarga, 17:16:27)

  karaf-4.2.3 bumps javax.annotation version to
1.3.0 (rovarga,
17:16:41)
  and multipatch fails of the dreaded package
resolution conflicts (rovarga,
17:19:08)
  we may need to postpone 4.2.4 upgrade until we
have sorted out oour JSR305 story, at least in APIs (rovarga,
17:19:28)


RFC8040 and filter (rovarga, 17:19:43)

  the question is whether the new RESTCONF
supports https://tools.ietf.org/html/rfc8040#section-4.8.4
filters (rovarga,
17:20:42)
  RFC8040 seems to specify filters only for
notification streams, not for individual GETs (rovarga,
17:26:34)
  since they are XPath based, we are pretty sure
they are not implemented for GETs even as an extension (rovarga,
17:26:54)


multipatch issues (rovarga, 17:27:22)

  the problem is with topic=fooBar, where the
latest tag is checked out and the patches are checked out on
it (rovarga,
17:28:54)
  for odlparent master patches cannot been
cherry-picked on top of odlparent-4.0.9 tag (rovarga,
17:29:12)
  https://jenkins.opendaylight.org/releng/view/integration/job/integration-multipatch-test-sodium/9/
(rovarga,
17:29:59)
  the best thing would be to introduce topic:foo
to perform a checkout (rovarga,
17:33:24)
  ACTION: LuisGomez
will try to come up with a solution for odlparent only (rovarga,
17:34:13)


tell-based protocol (rovarga, 17:36:10)

  ofp testing is still stuck on topology links
not being torn down (rovarga,
17:36:41)
  functional testing has only this issue
outstanding (rovarga,
17:37:46)
  https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-3node-clustering-perf-bulkomatic-only-neon/
(LuisGomez,
17:39:06)
  we have a failure with performance jobs, but
that may need tweaking (rovarga,
17:40:58)
  it seems that switching to tell-based protocol
by default is feasible in Sodium timeframe (rovarga,
17:45:04)


NETCONF scale testing (rovarga, 17:47:46)

  jmorvay asks about scaling of SB devices
(rovarga,
17:47:59)
  for a simple test tool, I could attach 10K
sessions with 965MiB retained heap (rovarga,
17:48:27)
  for stable/neon (rovarga,
17:48:37)
  with sshd-core-2.2.0 (staged at
https://git.opendaylight.org/gerrit/#/c/76300/) I could attach 10K
sessions with 744MiB retained heap (rovarga,
17:49:27)
  Bala reports similar results with 10GiB heap
and a real device (rovarga,
17:49:57)
  real device == ~70 YANG models (rovarga,
17:50:14)
  https://jira.opendaylight.org/browse/NETCONF-590 tracks
transport layer rework (rovarga,
17:53:02)
  which blocks
https://jira.opendaylight.org/browse/NETCONF-571 (rovarga,
17:53:38)
  which is about 50% of our current static
steady-state memory overhead (not counting SchemaContext
overheads) (rovarga,
17:56:28)


anyxml handling in get-config (rovarga, 17:57:47)

  Bala reports that it did not seem to work in
Carbon (rovarga,
17:57:59)
  it may be a missing bit in yangtools codecs,
where we do not handle one of the directions (rovarga,
17:58:30)
  will follow up after testing a