On 10/03/2022 06:39, Daniel de la Rosa wrote:

Hello Robert and all
On Tue, Mar 8, 2022 at 12:58 PM Robert Varga <nite@... <mailto:nite@...>> wrote:
On 07/03/2022 17:00, Daniel de la Rosa wrote:
> Hello TSC and all
>
> We are going to code freeze Sulfur for all Managed Projects ( cut
and
> lock release branches ) on Monday March  14th at 10 am UTC
>
> Please remember that we only allow blocker bug fixes in release
branch
> after code freezes
 From MRI perspective, we are ... 5? ... months behind the
schedule. The
good news is that MDSAL-696, which was blocking yangtools-8 -> mdsal-9
integration is done and dusted.
Next steps are to integrate with:
- controller-5 (done unless maven-verify finds something)
- aaa-0.15 (should be a breeze)
- netconf-3 (not trivial, but the pieces should be mostly there)
- bgpcep-0.17 (not sure, if we pass this we should be okay overall)
As it stands, I expect the MRI projects to be ready to integrate on
Friday the 11th, with bgpcep-0.17 integration gating yangtools-8
release.
great news, so sounds like we can still move forward with Sulfur code freeze on 9/14 ...

so a day has passed, I am up for an update:
- there was a regression (YANGTOOLS-1407) detected in lispflow integration, that is taken care of
- https://git.opendaylight.org/gerrit/q/topic:sulfur-mri is started, with most of the patches in a reasonable shape
- of that, there are some prerequisites in OFP which can be merged --
and Sangwook already merged most of them
- there is a model-vs-code impedance in OFP, I will looking into that tomorrow -- either the code needs to be adjusted or the model, both are quite easy to do
- there is the same problem in lispflowmapping, I think I know the answer, but fired an email at Lori to confirm
- both netconf and bgpcep builds are passing to ~50% mark

The major outstanding item are the MD-SAL regressions detected in ovsdb (and same thing is biting netconf) and bgpcep, tracked as MDSAL-731 and MDSAL-735 respectivelly.

I believe they are manifestations on of a problem with the new RuntimeType code -- and seem to have a good idea as to what needs to be done. I am currently working on a prototype. In terms of progress, it needs a bit of work for a test drive, but so far the code challenges are asking all the right questions, AFAICT.

If all goes well, I should have a fix ready tomorrow, but I won't know until it's complete enough to face testing.

Regards,
Robert

On 16/03/2022 00:53, Daniel de la Rosa wrote:

Hello Robert

Hello Daniel,

The major outstanding item are the MD-SAL regressions detected in ovsdb
(and same thing is biting netconf) and bgpcep, tracked as MDSAL-731 and
MDSAL-735 respectivelly.
Another day has passed :) so i guess we have to wait until these two are fixed to pick a RC?

Yes, unfortunately and I am still struggling with the fix, it turns out to be more complicated than expected, but I am converging on a solution.

Regards,
Robert

On 16/03/2022 23:46, Robert Varga wrote:

On 16/03/2022 00:53, Daniel de la Rosa wrote:

Hello Robert

Hello Daniel,

    The major outstanding item are the MD-SAL regressions detected in ovsdb
    (and same thing is biting netconf) and bgpcep, tracked as MDSAL-731 and
    MDSAL-735 respectivelly.


Another day has passed :) so i guess we have to wait until these two are fixed to pick a RC?

Yes, unfortunately and I am still struggling with the fix, it turns out to be more complicated than expected, but I am converging on a solution.

Alright, so MDSAL-731 is fixed and it makes MDSAL-735 more diagnosable -- I plan to squash that today.

Regards,
Robert

On 19/03/2022 16:37, Daniel de la Rosa wrote:

Alright, so MDSAL-731 is fixed and it makes MDSAL-735 more diagnosable
-- I plan to squash that today.
It looks like all issues have been resolved. So are we ready to pick the next AR as RC?

I am still working through netconf/bgpcep, plus the bump to MSI projects. I'll keep you posted.

Regards,
Robert

On 20/03/2022 19:50, Robert Varga wrote:

On 19/03/2022 16:37, Daniel de la Rosa wrote:

    Alright, so MDSAL-731 is fixed and it makes MDSAL-735 more diagnosable
    -- I plan to squash that today.


It looks like all issues have been resolved. So are we ready to pick the next AR as RC?

I am still working through netconf/bgpcep, plus the bump to MSI projects. I'll keep you posted.

bgpcep is ready for integration, but netconf is not -- as it turns the vast majority of patches dealing with SchemaNode.getPath() caller removal had to be reverted because ... let's say they were naive.

serviceutils is also ready to integrate, but there's a problem needing action in global-jjb: javadocs jobs are using mvn35 and thus end up exploding like here: https://jenkins.opendaylight.org/releng/job/serviceutils-maven-javadoc-verify-sulfur-openjdk11/32/

Even when we tell the job to use mvn38, it will unpack the correct maven, but will attempt to use mvn35: https://jenkins.opendaylight.org/releng/job/serviceutils-maven-javadoc-verify-sulfur-openjdk11/33/console

Regards,
Robert

Hello Robert,

On 20/03/2022 19:50, Robert Varga wrote:

On 19/03/2022 16:37, Daniel de la Rosa wrote:

    Alright, so MDSAL-731 is fixed and it makes MDSAL-735 more diagnosable
    -- I plan to squash that today.


It looks like all issues have been resolved. So are we ready to pick the next AR as RC?

I am still working through netconf/bgpcep, plus the bump to MSI projects. I'll keep you posted.

bgpcep is ready for integration, but netconf is not -- as it turns the vast majority of patches dealing with SchemaNode.getPath() caller removal had to be reverted because ... let's say they were naive.

Not yet for BGPCEP. I discover some late bugs on new PCE server last week. Patch is ready. I'll submit it this afternoon, just right after last verification.

Regards

Olivier

serviceutils is also ready to integrate, but there's a problem needing action in global-jjb: javadocs jobs are using mvn35 and thus end up exploding like here: https://jenkins.opendaylight.org/releng/job/serviceutils-maven-javadoc-verify-sulfur-openjdk11/32/

Even when we tell the job to use mvn38, it will unpack the correct maven, but will attempt to use mvn35: https://jenkins.opendaylight.org/releng/job/serviceutils-maven-javadoc-verify-sulfur-openjdk11/33/console

Regards,
Robert

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.

On 21/03/2022 14:45, Olivier Dugeon via lists.opendaylight.org wrote:

bgpcep is ready for integration, but netconf is not -- as it turns the vast majority of patches dealing with SchemaNode.getPath() caller removal had to be reverted because ... let's say they were naive.

Not yet for BGPCEP. I discover some late bugs on new PCE server last week. Patch is ready. I'll submit it this afternoon, just right after last verification.

Alright, anyway it's going to take a couple of days to sort out the RESTCONF stuff, so that (and the javadoc jobs) are the real blockers.

Regards,
Robert

On 21/03/2022 14:24, Robert Varga wrote:

On 20/03/2022 19:50, Robert Varga wrote:

On 19/03/2022 16:37, Daniel de la Rosa wrote:

    Alright, so MDSAL-731 is fixed and it makes MDSAL-735 more diagnosable
    -- I plan to squash that today.


It looks like all issues have been resolved. So are we ready to pick the next AR as RC?

I am still working through netconf/bgpcep, plus the bump to MSI projects. I'll keep you posted.

bgpcep is ready for integration, but netconf is not -- as it turns the vast majority of patches dealing with SchemaNode.getPath() caller removal had to be reverted because ... let's say they were naive.

So fixing RESTCONF here is going to require a yangtools-8.0.2 release. The problem is that we lost a DataSchemaContextTree capability which was provided indirectly via SchemaNode.getPath() -- the details are in YANGTOOLS-1412.

I expect to have that fixed today and finish the RESTCONF integration tomorrow, so that the MRI bump should (fingers crossed) over the weekend.

It is way later than planned, expected or hoped, but unfortunately it is what it is. There is silver lining here, though: this effort found two bugs that have been squashed after lurking around for 7 years.

I'll keep you posted,
Robert

--

Hello Robert, all,

On 21/03/2022 14:45, Olivier Dugeon via lists.opendaylight.org wrote:

bgpcep is ready for integration, but netconf is not -- as it turns the vast majority of patches dealing with SchemaNode.getPath() caller removal had to be reverted because ... let's say they were naive.

Not yet for BGPCEP. I discover some late bugs on new PCE server last week. Patch is ready. I'll submit it this afternoon, just right after last verification.

Latest patches have been submitted:

• Correct bugs in Path Manager: https://git.opendaylight.org/gerrit/c/bgpcep/+/100205/4
• Add Close Loop mechanism to PCE Server: https://git.opendaylight.org/gerrit/c/bgpcep/+/100206/5
• Add Exclude and Include Route for Path Computation: https://git.opendaylight.org/gerrit/c/bgpcep/+/100247/1

The first two have been reviewed and are ready to be merged. Latest one needs to be reviewed.

The only point that is remaining for BGPCEP, is to clean the compilation from various warning message. I don't know if we include these now or wait for Sulfur SR1.

Regards

Olivier

Alright, anyway it's going to take a couple of days to sort out the RESTCONF stuff, so that (and the javadoc jobs) are the real blockers.

Regards,
Robert

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.

On 24/03/2022 01:24, Daniel de la Rosa wrote:

On Wed, Mar 23, 2022 at 3:47 PM Robert Varga <nite@... <mailto:nite@...>> wrote:
That’s impressive. Ok thanks and it is all good. We will have a very solid sulfur release

A quick update here. We are on the home stretch now, with 31 callsites still being bad. About a third are duplicate (e.g. they are the same weirdness across bierman20/rfc8040). They will need some amount of TLC, which I am not in position to give right now, sorry :(

I expect to have them cleared tomorrow, Tuesday the latest, at which point I'll activate the supercommitter rights and roll things out.

The upgrade guide will come right after that, e.g. Tue or Wed.

Regards,
Robert

On 28/03/2022 00:10, Robert Varga wrote:

On 24/03/2022 01:24, Daniel de la Rosa wrote:

On Wed, Mar 23, 2022 at 3:47 PM Robert Varga <nite@... <mailto:nite@...>> wrote:
That’s impressive. Ok thanks and it is all good. We will have a very solid sulfur release

A quick update here. We are on the home stretch now, with 31 callsites still being bad. About a third are duplicate (e.g. they are the same weirdness across bierman20/rfc8040). They will need some amount of TLC, which I am not in position to give right now, sorry :(

Okay, so I am down to 3 call sites, which fall into two categories:

1. old RESTCONF's URI parser leafref support. I think I can perform sufficient surgery to make this work without having to rewrite that thing (as was done for new RESTCONF, where this is a breeze)

2. two instances of a weird interaction with XML codec. It is not that weird, I understand what is going on, it just needs a bit of more work.

I will need another day to deal with them and, fingers crossed, UTs should agree.

Regards,
Robert

On 29/03/2022 01:13, Robert Varga wrote:

On 28/03/2022 00:10, Robert Varga wrote:

On 24/03/2022 01:24, Daniel de la Rosa wrote:

On Wed, Mar 23, 2022 at 3:47 PM Robert Varga <nite@... <mailto:nite@...>> wrote:
That’s impressive. Ok thanks and it is all good. We will have a very solid sulfur release

A quick update here. We are on the home stretch now, with 31 callsites still being bad. About a third are duplicate (e.g. they are the same weirdness across bierman20/rfc8040). They will need some amount of TLC, which I am not in position to give right now, sorry :(

Okay, so I am down to 3 call sites, which fall into two categories:
1. old RESTCONF's URI parser leafref support. I think I can perform sufficient surgery to make this work without having to rewrite that thing (as was done for new RESTCONF, where this is a breeze)
2. two instances of a weird interaction with XML codec. It is not that weird, I understand what is going on, it just needs a bit of more work.
I will need another day to deal with them and, fingers crossed, UTs should agree.

This is now cleared, netconf-3.0.0 is out there. Most of the MSI update patches are done -- except openflowplugin.

Upgraded spotbugs is finding some badness which was introduced in https://git.opendaylight.org/gerrit/c/openflowplugin/+/91313 and I need to figure out how to fix it.

BGPCEP still has a four contributions left for review, I will take care of that tomorrow.

Regards,
Robert

On 31/03/2022 01:20, Robert Varga wrote:

On 29/03/2022 01:13, Robert Varga wrote:

On 28/03/2022 00:10, Robert Varga wrote:

On 24/03/2022 01:24, Daniel de la Rosa wrote:

On Wed, Mar 23, 2022 at 3:47 PM Robert Varga <nite@... <mailto:nite@...>> wrote:
That’s impressive. Ok thanks and it is all good. We will have a very solid sulfur release

A quick update here. We are on the home stretch now, with 31 callsites still being bad. About a third are duplicate (e.g. they are the same weirdness across bierman20/rfc8040). They will need some amount of TLC, which I am not in position to give right now, sorry :(

Okay, so I am down to 3 call sites, which fall into two categories:

1. old RESTCONF's URI parser leafref support. I think I can perform sufficient surgery to make this work without having to rewrite that thing (as was done for new RESTCONF, where this is a breeze)

2. two instances of a weird interaction with XML codec. It is not that weird, I understand what is going on, it just needs a bit of more work.

I will need another day to deal with them and, fingers crossed, UTs should agree.

This is now cleared, netconf-3.0.0 is out there. Most of the MSI update patches are done -- except openflowplugin.
Upgraded spotbugs is finding some badness which was introduced in https://git.opendaylight.org/gerrit/c/openflowplugin/+/91313 and I need to figure out how to fix it.
BGPCEP still has a four contributions left for review, I will take care of that tomorrow.

Alright, everything has been merged up.

int/dist and autorelease need to have their jobs updated to use maven-3.8, please review the patches here:

https://git.opendaylight.org/gerrit/c/releng/builder/+/100304
https://git.opendaylight.org/gerrit/c/releng/builder/+/100306

I am writing up the release notes and upgrade guide next.

Bye,
Robert

On 31/03/2022 17:41, Robert Varga wrote:

Alright, everything has been merged up.
int/dist and autorelease need to have their jobs updated to use maven-3.8, please review the patches here:
https://git.opendaylight.org/gerrit/c/releng/builder/+/100304
https://git.opendaylight.org/gerrit/c/releng/builder/+/100306
I am writing up the release notes and upgrade guide next.

So looking at https://jenkins.opendaylight.org/releng/job/integration-distribution-test-sulfur/185/ we have plenty of badness.

The core problem is that a ton of test suites are depending on OPERATIONAL_API, which points to old RESTCONF, which is no longer installed by default. I have https://git.opendaylight.org/gerrit/c/releng/builder/+/100359 to install it for now, but we need to eventually (very soon) ditch it.

Luis, any idea on how to fix that in a sensible manner?

Also, RF is now reporting a ton of ugly 'Keyword 'BuiltIn.Run Keyword Unless' is deprecated.' messages -- I'll take a peek next week unless somebody beats me to it.

Finally, I am tracking two regressions in MD-SAL (already fixed) and one in NETCONF (next on the list), but I'd like to get a better reading on CSIT before rolling out the corresponding releases out.

Regards,
Robert

On 06/04/2022 01:21, Daniel de la Rosa wrote:

Hello Robert and all
it looks like we still have several issues
https://jenkins.opendaylight.org/releng/view/autorelease/job/integration-distribution-test-sulfur/190/ <https://jenkins.opendaylight.org/releng/view/autorelease/job/integration-distribution-test-sulfur/190/>
are we still tracking some of these issues in jira tickets?

Yes, I think most of the issues are already fixed, but I am waiting for https://jira.linuxfoundation.org/plugins/servlet/theme/portal/2/IT-23826 to issue the corresponding releases.

With the upgrade guide complete, I think we are good to branch, though.

Regards,
Robert