2024 meeting minutes

Meetings connection details

available at TransportPCE meetings
https://lf-opendaylight.atlassian.net/wiki/display/ODL/TransportPCE+meetings

2024-11-21

General Information

We still need to update the revision notes for Sc SR1.

Code Review

No merged change this week.

Changes transportpce under review on master branch:

“Handle modifications in existing RPCs and new RPCs”, “Create topo Context at Init for Alien & SBI”, “Add Tapi Abstracted Node to OR Topo” and “Add Func Test for Topology extension” pass the gate and are ready for merge after a clean-up.
Relation Chain of Smartoptics :
- “Consistent cent. freq. width decimal precision” Corrects an issue where the nr of decimals for central frequency was 4 in ZtoA direction and 5 in AtoZ direction
- “Implement service-format 'other'” : This update introduce a new service-format which ensures the mc interface is 8GHz wider than the nmc interface.
- “Implement client create service frequency choice” : Implement the augmentations found in transportpce-service-spectrum-constraint@2023-09-07.yang. The opportunity for this change was discussed during the meeting and it revelency needs to be investigated.
- “New packages simplifying picking service frequency” : simplify selecting an available frequency range from which a subset may be setup for a new service. It has been reviewed by Olivier and LGTH.
- “Augment service-create adding frequency properties” : the yang file has be renamed from com-smartoptics-openroadm-service.yang to transportpce-service-spectrum-constraint. This change is ready to be merged unless there are clarifications to be made on ranges of values to be specified in the model.
- “Check if SRG has available PPs before access” : If all PPs are used for any SRG, node validation will crash in getRdmSrgClient because it tries to access the first element of the empty available PP set, causing path computation failure even if there are other SRGs with available PPs. This fixes the issue by moving the existing check for empty available SRG PPs to before trying to access the first element. The change is ready for review and passes the gate.
- “Ignore end node without available TPs in PCE” corrects a bug so that a node that has no available tps is effectively ignored. It is ready for review and passes the gate
- “Set ServiceFeasibilityCheck to false for some rpcs” from Jonas has been rebased and passes the gate. The change has been reviewed by Olivier. If Christophe review is positive, this change could be merged.
“Add CEPlist to OTS NEP at init” : Add Cep list to Owned node edge points for Xponders at initialization corresponding to T0 configuration. This change allows completing the topology as required for the purpose of debugging the relation chain associated with the PCE link and node classes.
“Fix ROADM provisioning for 7.1 models ” : Bala has added his signoff so that the change passes the pre-commit, but asks not to merge the change since additional work is needed. Especially when a 2.2.1 ROADM is connected first and a 7.1 connected after, the neighbor is not recognized and the links between them are not created. Bala needs to update some codes in the OLM to implement for 7.1 span-loss retrieval.

Other items

Olivier briefly introduced the way topology extensions are managed

2024-11-14

General Information

Scandium SR1 has been released the 11^th of November. We still need to update the release notes.

Code Review

3 changes merged during the week : “Adjust tox ini tests sequencing”, “Yang file to augment OR Topology”, and “networkutils RPCs adjusted for topo extension”. This corresponds to the bottom relation chain on TAPI work.

Changes transportpce under review on master branch:

“Handle modifications in existing RPCs and new RPCs” passes the gate and is ready for merge.
“Add Tapi Abstracted Node to OR Topo” : still does not pass the gate but is probably due to an issue on the gate (221 tests).
“Add Func Test for Topology extension” : still does not pass the gate but is probably due to an issue on the gate (221 tests).
“Uninstall tapi feature at the end of functional tests” is a standalone change that could be merged on other branches than master (on master branch, the code it is already integrated in “Add Tapi Abstracted Node to OR Top”). The purpose of the change is to uninstall the tapi feature at the end of each tapi functional test.
“Migrate to Apache Pekko” : allows to use Pekko in place of Akka. This change does not pass the gate. Robert has been added as a reviewer to help understanding us why.
Relation Chain of Smartoptics : The changes at the bottom of the stack have been merged. Joakim shall rebase next changes on the master branch.
“Augment service-create adding frequency properties” : the yang file shall be renamed from com-smartoptics-openroadm-service.yang to something like transportpce-service-spectrum-constraint.yang : including the name of an OEM in the title of the yang file suggests that this file is dedicated to specific equipment, which is not the case. Also having openroadm in the title could bring confusion as this is an extension of openroadm models but is not part of them.
“Set ServiceFeasibilityCheck to false for some rpcs” from Jonas shall be rebased to check that it passes the gate.
“Fix ROADM provisioning for 7.1 models ” : Bala shall add his signoff so that the change passes the pre-commit.

2024-09-19

General Information

Scandium is not released yet because of some regressions detected in the NETCONF project. So we are waiting for the availability of the latest NETCONF dependency. Robert published this afternoon netconf-8.0.2 which corrects the bug we reported for the swagger issue.
Ca-SR2 release is also pending for the same reason.

Code Review

No change merged during the week

Changes transportpce under review on master branch:

Olivier pushed the relation chain associated with Node and Link Classes built for the T-API PCE on master, and set to abandoned more or less outdated corresponding changes on stable/calcium. This includes 4 changes, the initial one “Initial Tapi-2.1 node and link classes for PCE” focusing on 2.1 model that will never pass the gate since it does not compile (using unsupported Tapi models). The second , “Initial T-API 2.4 PCE link & Nodes Classes" that provides the classes based on current T-API model 2.4. “Add Tapi Topo File for test" provides the topology file that will be used for tests. Last one “Add Junit tests of Tapi link & Node classes” provides the test associated with these classes. All these changes are defined as WIP. They will be set as Active as soon as the test files have been refreshed with the latest version that allows going through all Junit tests.
“Update portmapping model” should be abandoned. This change was pushed by Olivier but was done making a wrong assumption about the supporting-odu4 is configured in PortMapping and used later on. “Corrections of 121 tests (portMapping update)" shall also be set as abandoned.
“Populate OM in portMapping and Topo” could be merged as soon as someone review it and potential adaptations are performed.
“Fix a few bugs processing Topo” could be merged as soon as someone review it and potential adaptations are performed.
“Link node Id to node name conversion in TAPI” has been submitted by Joakim. It Corrects an issue where a value such as ROADM-A-xxxxx would not return the node name ROADM-A. Joakim integrated the comments.

2024-09-12

General Information

Scandium shall be released next week. Gilles released tpce artifacts. But some regressions were detected in the NETCONF project. So we will need to take into account the new NETCONF version.
Gilles also started the release process for Ca-SR2 but NETCONF project has the same impact.
Gilles raised a Jira Ticket NETCONF-1386 to highlight the fact that the swagger interface is again not working.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

3 changes related with the creation of Scandium Branch : “Update .gitreview…” & “Bump upstream dependencies …”.
Same for Ca-SR2 branch
The fix provided by Joakim have been merged on both master and Ca branches.
Changes associated with lightynode Upgrade.

Changes transportpce under review on master branch:

Gilles reworked the relation chain of Olivier, squashing “modify imports to use last version of PortMapping” in change 113427
Olivier made some corrections to “Update portMapping models” which allows passing 221 & tapi tests locally. However he stills experience some issues with 7.1 and hybrid tests, and requires help since the issues do not seem to be directly related to the adjustment.
Nothing new this week on other changes.

2024-09-05

General Information

Scandium shall be officially released in 2 weeks. Tpce Scandium migration is almost done.
CA-SR2 is almost 1 month late. Gilles backported some changes on the stable/calcium branch. He is experiencing a lot of CI issues, mostly with maven-verify job. The root cause could be on the aaa project. It seems that, among other things, passwordService dependency is missing for org.opendayligth.aaa.datastore.h2.H2Store bundle.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

“Increase the Kafka producer reconnect backoff time”: configures the kafka producer with a reconnection time at 10min (instead of 500ms) to avoid polluting the logs with too many warnings messages when no Kafka server is available (master and backported on stable/calcium branch).
“Improve lightynode installation process”: openroadm-device is installed only once for all tests
“Bump upstream dependencies for 2024.09 Scandium”. Adopts latest dependencies for odlparent & mdsal (tpce/models)

Changes transportpce under review on master branch:

“Bump upstream dependencies for 2024.09 Scandium” (tpce),is the scandium migration. It adopts latest dependencies for odlparent, mdsal, yangtools, Netconf & tpce models and disables buildlighty from tox voting. It could (and shall be) merged asap.
Smartoptics relation chain: Olivier reviewed the change. It could be merged after being rebased on scandium.
“Migrate to Apache Pekko”: shall allow to use Pekko instead of Akka. Ivan warned that Pekko will in fact be used only in Titanium. So this shall not be merged rigth now.
“Change supporting-odu4 to leaflist in portMapping”, “Modify imports to use last version of portmapping”, and “Populate OM in portMapping and Topo” provides some enhancement in PortMapping and allow to populate from device (7.1) operational modes in portMapping and propagate them in the topology.
changes related to the IPoWDM use case: no new updates since last week.
“Upgrade lightynode to Ca-SR1”: now that scandium is OK, could be merged.

Changes transportpce under review on stable/calcium branch:

Some changes have been backported, but some tests do not pass, and maven-verify systematically fails on the different feature tests
The changes concerning TAPI are in a different states than what has been pushed on master. This will need to be solved and clarified.

2024-08-29

General Information

ODL's TSC has just been renewed: Technical Steering Committee (TSC) - OpenDaylight - Confluence (atlassian.net)
Code Freeze for Scandium happened 2 weeks ago. Master branch shall be migrated to Titanium. The ODL Scandium release is planned on the 19^th of September.
Our ODL JIRA and Wiki have been migrated to Altassian Cloud. For Jira a double authentication is needed. Some convenient features have been lost (link with Gerrit change associated with the Jira Ticket not yet effective).
LFN is thinking moving the tox-verify jobs from Jenkins to Github Action

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

“Search available frequencies in TAPI causes errors”: fix a bug in TAPI module
“Fix Shellcheck in InstallMavenUbuntu script”: minor improvment in shell script
“Increase the kafka producer reconnect backoff time”: configures the kafka producer with a reconnection time at 10min (instead of 500ms) to avoid polluting the logs with too many warnings messages when no Kafka server is available. This change merged on master has been cherry picked and stable/calcium branch, but the CI is failing
- it brings up a new problem related to the installation of lightynode simulator on the CI. The lightynode installation process should be improved since the same simulator version is used for the different org-openroadm-device.yang model version tests

Changes transportpce under review:

changes related to the IPoWDM use case and just upadated by Bala. They are now in review:
- “Add common-id to path-computation”: OK for merging, but before add a Jira ref
- “Read optical-operational-mode in temp-service-create” allows retrieving the op-modes provided in the requesting-interface-properties container of temp-service-create, which is needed for path computation with external pluggable. It currently fails at the service handler level. JUnit tests of servicehandler still need to be updated with code modification done.
- “Add OSNR calculations for IPoWDM use case” provides impairment aware path computation according to the operational modes provided in the temp-service-create in IPoWDM use case. Olivier shall review this change.
One change from Ivan Hraško on migration from akka to Apache peko:
- “Migrate to Apache Pekko”: we need to wait for change111858 on controller project to be merged before merging it.
changes on TAPI module proposed by Smartopics:
- “Frequency rounding error in GridUtils” is a new bug detected by smartoptics. At first sight, we are ok with this fix
- "Error due to empty list of XPDR in TAPI", "Error due to duplicate roadm node id key in TAPI", "Error in TAPI creating frequency ranges" are other bugs detected in TAPI. OK with that, but they are above the following change that still need a review...
- "New Tapi package dealing with frequency BitSets": proposes a new TAPI package dealing the frequencyBitSets. After exchange between Joakim and Olivier, this package has been refactored by Joakim… Need to be reviewed again by Olivier (and others)
“Upgrade lightynode to Ca-SR1” : Gilles prefers to merge this change after major issues related to scandium migration have been fixed in order to have a stable simulator version for tests
“Bump Upstream dependencies for 2024.09 Scandium”: Gilles started the migration of master branch to scandium. The compilation with JUnit tests is OK, but lots of functional tests are failing for two main reasons:
- the return of the request when the requested data in the datastore does not exist changed
- the netconf device notification registration is no longer effective with lightynode simulator (it works with honeynode, but honeynode is no longer used for TPCE functional tests on the CI)
- Gilles reported these issues towards upstream projects. Waiting for their feedbacks...

2024-07-25

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

No change merged since last meeting.

Changes transportpce under review:

“Bump CI scripts to Ubuntu 20.04”, at the bottom of an important relation chain, is dedicated to the migration of CI runners. This topic was relaunched this morning and a change was pushed to releng project. Even if the proposed solution may not be the best way to solve the issue, Gilles may need to merge it, because a lot of changes are on top of it.
“Search available frequencies in TAPI cause error” has been reviewed and could be merged.
“New Tapi package dealing with frequency BitSets” change has been commented by Olivier. He feels that this change brings abstraction and is quite complex to analyze, while being based on a fixgrid approach, whereas transportPCE has evolved towards a Flexgrid approach. He will review the answer of Joakim to his comments
“Error in TAPI creating frequency ranges” has still not been reviewed since it leverage the previous change, on which we have some concerns.
“Error due to duplicate roadm node id key in TAPI” is independent from previous changes, solve some bugs and has been reviewed: it could be merged.
“Change supporting-odu4 to a leafleast in portMapping”: Olivier pushed this change because the support of OUT/ODUCN implies to have several ODU4 interfaces on top of an ODUCn one.
“Populate OM in PortMapping and topo”: this changes proposes to populate from the device (only R7.1) the supported operational modes in portMapping, and to propagate the PortMapping corresponding list in the topology.
“Fix a few bugs processing Topo” has been pushed by Olivier. While debugging TAPI PCE abstraction classes, it appeared that the topology used for the test is uncomplete. It becomes obvious that some modifications were needed in the process of building the topology. This change is applying some correction to the process.
“UpgradeLightyNode to Ca-SR1” should be ready for a merge.
“Read optical-operational-mode in temp-service-create” pushed by Bala is still Work in Progress.

2024-07-18

General information

Migration to Scandium shall start soon. Scandium will require Java 21 and Maven 3.9.5 for build and run.
The gate is in bad shape!

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

No change merged since last meeting.

Changes transportpce under review:

“Add common-id to the path computation request” & “read optical-operational mode in temp-service-create” are 2 changes pushed by Bala to handle the IPoWDM use case.
Olivier has pushed a number of Change related to T-API, mainly for backup, all of them being in WIP state. The changes pushed on stable/Ca will need to be pushed on Master for simplification of the integration of these changes.

Other

Discussion about IPoWDM Use case:
- The user can push more than one mode in the temp-service-create. A priority is associated to them. Post algo validator is responsible for evaluating the impairment on the path and compare them with the specifications of the transponder, determined by the operational-mode. For tunnel, at that time, _and because the code was written before we model the IPoWDM use case in OpenROADM_ PostAlgoValidator does not compare OSNR resulting from the degradation on the OLS with the performance of end-terminals, since we had no way to know those.
- Now that we use models allowing the implementation of IPoWDM Use Case, the code of the PostAlgoValidator for tunnels shall be modified to do exactly what is done in a regular case. The only difference is that the performances of the end-terminals are not extrapolated from the last node of the path (because in the case of a tunnel, this is a SRG Add/drop, and not a transponder) but from the operational-modes that have been provided in the container “requesting-interface-properties” of the temp-service-request. But the calculation and the way performance are retrieved in the catalog from the operational-mode do not change.

2024-07-11

General information

ODL scandium release (planned for September) is underway in upstream projects: wil require Java-21 for build and runtime, and Maven-3.9.5+

Code Review

Changes transportpce under review:

Smartoptics proposed a relation-chain including several changes dedicated to TAPI and more specifically some issues associated with the handling of frequency ranges and bitset.
“Bump CI script to Ubuntu” shall revert and repair the CI, but needs to be adjusted to use maven environment variables.
“Upgrade LightyNode to Ca-SR1” provides a new version of lighty aligned with lighty-core-20.1.0.
Bala pushed some changes associated with IPoWDM use case: “Add common-id to path-computation-request ” and “Read optical-operational-mode in temp-service create”. This last shall allow handling more than just one supported-operational-mode, and select the most appropriate one, based on the associated preference criteria.

2024-07-04

No Meeting. Independence day in US.

2024-06-27

General information

Ca-SR1: has been approved by the TSC so that it shall be released in a short time

Code Review

Changes transportpce under review:

“Address Comments on Change 111419”: the title shall be independent from any reference to Gerrit.
“Deleting openroadm after controller restart” Olivier made a review on Jonas’ change.
“Bump CI scripts to Ubuntu 20.04”. Gilles pushed this change to repair the gate.

Other

Christophe has identified that the issue with lighty node comes from the fact that a bug in Yang-Tools for models that have complex interactions, solved in version 13.0.4 of Yang-Tools, available in Ca SR1, has still not been integrated in Lighty. Christophe would like to push OpenConfig R2.0 models on transportpce. No objection.

2024-06-20

General information

Ca-SR1: is still under TSC validation
K-SR3 has been released.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

“DebugTool for openconfig proprietary extensions ”
All changes related to the re-spin of Ca SR1
- Bum netconf 7.0.7, project versions
- Fix CI CentOS8 isue: workaround from Guillaume. Revert a commit to reenable voting for checkbashisms an pre-commit after CI has been migrated to Ubuntu 2004
- Their backport to master

Changes transportpce under review:

“Bump CI scripts to Ubuntu 20.04 ” modifies scripts to select correct version of Java (17) so that our test run correctly on the CI now that it is migrated to Ubuntu. However a lot of changes are still failing at the maven-verify-job step. The issue was raised to Robert. Gilles waits for an answer prior to merge this change.
“Address comments on change 11439” addresses the remaining comments made by Guillaume.
“Deleting openroadm node after controller restart”: status did not change. Nobody had time to test it on real equipment.

Other:

Christophe is working on migration of LightyNode to align with Ca SR1 distribution. He managed to handle it with OpenROADM models, but did not succeed in making it work with OpenConfig models.

2024-06-13

General information

Ca-SR1: shall be under TSC validation, but a new issue has been identified in the NETCONF project. We could need to respin it...
K-SR3: Gilles released TPCE artifacts today morning. The release candidate shall be under TSC validation today.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

All changes related to the re-spin of Ca SR1
“Fix floating equality bug in tapi module” corrects a bug: == cannot be used to compare 2 floating values
“Make tox pre-commit and checkbashism non voting”: the linux image used on the gate (centOS 8) was end of builds. To solve temporarily this issue, this change was merged.
“Fi x CI Cent OS 8 configuration issue (workaround)” has been proposed by Guillaume and merged after to solve temporarily in a better way this issue.
“Fixup checkstyle issue” solves some new code style issues raised by Spotbug.
Some of the changes merged on Ca branch associated to TAPI implementation were backported to the master branch. For this last, lighty could now be reactivated on the gate.
“Bump lighty to 20.0.0” was also backported to the master branch. Lighty could now be reactivated on the gate for our functional tests.
“Consolidate connectivity Utils” shall be corrected to address Guillaume changes and pushed to master branch.

Changes transportpce under review:

“Debug tool for OpenConfig proprietary extensions” is almost ready to be merged. Guillaume would appreciate some feedback from the IOWN community.
“Deleting openroadm node after controller restart” from Jonas need to be rebased and shall be tested prior to be merged.

2024-06-06

General information

Ca-SR1 shall be under TSC validation, but an issue has been identified (One dependency update on the AAA project was forgotten). Ca SR1 release process need to be respined now.
K-SR3 should be ready by 2 weeks.

Code Review

No change merged during last week.

Changes transportpce under review:

“Fix frequencies in TAPI convertORToTapiTopology” will be integrated in the respin of Ca SR1 release
We are in an intermediate phase since Gilles had to prepare Ca SR1 during his vacations and did not have the time to backport on master some the changes that were pushed on stable/calcium branch. He will start working on backporting required changes on master branch in coming week. For what concerns changes associated to the TAPI feature, we will need to be careful, since some of them have been pushed on Ca branch, whereas some others were pushed on master branch.
“Deleting openroadm node after controller restart”: Guillaume made a review, but Jonas is probably expecting a review from people mastering operational aspects prior to apply required modifications, especially as some functional tests are impacted.
“Debug Tool for openConfig proprietary extensions” is according to Guillaume ready to be merged.

2024-05-23

General information

Ca-SR1 is under progress: Gilles backported from master all changes of interest for the release. An issue detected by our functional test has been solved thanks to Pantheon regarding Lighty.io, but we still experience lots of issues with the gate, due to several cumulative issues on the infrastructure. Need to relaunch everything...
K-SR3 should start very soon.
Scandium is planned for September.

Code Review

Changes transportpce under review:

Jonas is asking for feedback for his change “Deleting openRoadm node after controller restart”. We invited him connecting to the next public meeting.
Olivier’s last changes with regard to T-API have been initially pushed on Ca, whereas Guillaume refactoring was made on the master branch. At the end backporting the refactoring on the Ca branch created multiple conflicts that were not so easy to handle. After several tentative the T-API change have been rebased, but we needed to step back on a few functions and went back to initial code to have tapi functional and Junit tests passing.

2024-05-16

General information

Ca-SR1: tpce artifact will need to be published next week as soon as the Netconf dependency is available.
K-SR3 should have been released last week, so it will come soon.
Scandium is planned for September.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

“Remove bad file from OpenConfig Module” on transportpce-models: mistake when openconfig models had been added to the repo
“Update release in docs/conf.yaml” will also need to be updated on other stable branches
“Abort power setup if setting gainloss fails” has been merged by Guillaume once “Power setup gainloss failure Junit test” provided a unitary test to check the correct behavior of previous change.
“Bump transportpce-models upstream dependency”: explain why the gate was failing for a while after Ca was released
“Refactor CovertORTopoToFullTopo Step 1 to 7”: Guillaume refactored and factorized the code for this test since it has been modified from initial version by different coders and needed some consolidation.
“Refactor TAPI Utils TapiContext”, “Refactor TAPI TapiNetworkModelServiceImpl”, “Refactor TAPI TapiNetworkModelServiceImpl code style” and a few others, improve indentation and provide other improvements.

Changes transportpce under review:

Gilles worked on a few changes around dependencies update and project version bump to address Ar SR4 but will be abandoned since this release in fact only concerns core projects.
“Consolidate ConnectivityUtils" associated with “refactoTapi2.4 Final touch” bring the latest modifications to the migration to T-API 2.4 adding some unitary tests to check functions that have been added to handle Photonic Media NEP spec and the parameters associated to the OMS.
“Initial T-API 2.4 PCE link & Nodes Classes” daisy chained with “Initial Tapi-2.1 node and link classes for PCE” is the old change that provided the link and node classes required for the PCE to operate from a T-API topology. It has been rebased on master branch. All associated code needs to be tested using a topology file that will be generated as soon as we migrate to Ca-SR1 that re-activates the swagger function required to retrieve the full topology from the Data Store.

2024-05-09

No Meeting. Public holiday.

2024-05-02

Code Review

Changes merged since last meeting on transportpce:

Guillaume added a perl linter

Changes transportpce under review:

Gilles pushed few changes on stable/argon branch to prepare Ar-SR4 (last release for Argon release train), but unfortunately jenkins jobs have been removed in releng/builder...
new change on master to update the transportpce-models version (the gate was no longer working)
several refactoring changes from Guillaume on TAPI code: few of them are ready to be merged

2024-04-18

Meeting cancelled

2024-04-11

General information

LFN DTF summit will be hold on the 2^nd and 3^rd of May in San Jose. Deadline for registration is on the 12^Th of May.
We are using Teams for tpce meetings, and the LFN recommends to use their Zoom tool. However neither Orange nor AT&T company authorize to use zoom tool...
Gilles published last week TPCE artifacts for Ca. For what concerns tpce, everything has been done. The distribution shall be released next week after the TSC as approved it.
A change in Netconf is currently under review to solve our OpenAPI issue.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

2 changes concerning the Ca release have been merged.
Gilles removed the tapi models 2.1.1 from transportpce-models which are not used anymore.
The set of Openconfig models has been backported to stable/potassium branch and will be included in K-SR3

Changes transportpce under review:

Changes corresponding to the branch managements, bump of project versions associated to Ca and the evolution of the different branches.
Note : We have diverged so much from K in Ca that only a few changes will be backported to K for SR3 release.
Guillaume changes associated to the refactoring of tapi module as well as the correction of 2 bugs that were reintroduced through wrong multiple merge-conflicts management.
“Use Map in TAPI topology ConvertORToTapiTopology” is a code optimization proposed by Guillaume to reduce the cyclomatic complexity
“Update tpce release note for Calcium” has been merged, but we forgot to mention the Rollback -process refactoring of SmartOptics and the AT&T work around IPoWDM use case. Will be updated.

Other

sprint 34 is cloed in session, and the next one launched.

2024-04-04

General information

Ca GA is still under progress and should be available very soon now. TPCE is more or less the only project that needs to be released now.
Transportpce has been demonstrated last week during the OFC event. Bala provided some feedbacks. A lot of people followed the event this year and there was a lot of visitors on OpenROADM booth

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

Some changes related to the Ca preparation: “Refactor NetworkModel for PortMappingListener”, “Replace javax.activation-api by the jakarta one”, “Bump grizzly-http-server version to 4.0.2, “Bump netconf to version 7.0.4”, , “Bump netconf to version 7.0.3”, “Fix karaf client issue in functional test library”, “Fix bugs in nbinotifications module”, “Migrate tapi notification functional tests to 2.4”, “Bump kafka-client version to 3.7.0”, “Bump Netcof Version to 7.0.4”
In addition, a set of changes from Guillaume associated with refactoring of both Ca associated changes and Tapi Refactoring

Changes transportpce under review:

Olivier is currently working on handling translation of OMS attributes of OpenROADM links into the NEPs and CEPS that end tapi links. This implies to some refactoring of Tapi Modules to have the need function in the best place and avoid redundancy.

2024-03-21

General information

LFN migrated its servers from US to Europe: as a result, JIRA, Gerrit and CI/CD experienced some issues that have been rather solved.
Ca should have been released on the 14^Th but it is still under preparation.

Code Review

Changes merged since last meeting on transportpce-models and transportpce:

“Bump of upstream dependencies to latest Ca versions” and “Bump Netconf to 7.0.3”: let's notice that a new dependency for Netconf shall be available soon.
“Make lighty build non voting in CI” disactivates Lighty runtime build since lighty-core has not been migrated to Ca yet.
“Fix Renderer Junit Tests exceptions” : the change from Joachim had some impact on the tests and Gilles had to reworked some Junit tests.
“Only verify PreFec Ber if end node is Xponder”.
“Refactor common CrossConnectImpl” and “Fix inventory listeners code style”.
“Refine the RPC implementation registration” replaces the ClassToInstanceMap argument by a plain varargs with the RPC implementations when registering RPC to mdsal.binding.api.RpcProviderService. This change also allows optimizing code by removing classes that have now become almost useless. JUnit tests were also adapted to check exactly which RPC impl are registered.

Changes transportpce under review:

“Refactor NetworkModel for PortMappingListener” has been proposed by Guillaume and is ready to be merged, but after Gilles relation chain for the migration to Ca is done.
“Replace javax.activation-api by the jakarta one” removes the fixed and old version of javax.activation-api for the benefit of the jakarta.activation-api provided by odlparent.
“Bump grizzly-http-server version to 4.0.2" replaces old and unsecured 2.4.0 version .
“Bump netconf to version 7.0.3”: this version of Netconf was supposed to solve Swagger issues, but it still not the case for NB API
“Fix karaf client issue in functional test library”: Removes logout from the bash command sent with subprocess run to avoid having a runtime exception in karaf when we install the odl-transportpce-nbinotifications feature and removes a useless variable in sims. It also updates_cp_port_ntcf method .
“Fix bugs in nbinotifications module” fixes NPE issue in NbiNotificationsHandler, message issue in PceNotificationHandler of service handler, adds small delay in functional test execution to let the notification being sent in kafka broker and fixes issue when uninstalling the feature (publisher sessions with kafka server remained open)
“Migrate tapi notification functional tests to 2.4” fixes NPE issue in CreateConnectivityServiceValidation when the request does not contain any topology constraints, and also completes TAPI notification functional test suite. It adapts the tests to the new implementation.
“Bump kafka-client version to 3.7.0” allows to have now all dependencies aligned with current version.
“Refactor a few Renderer RPCs” focuses on code style and some optimizations.

2024-03-14

General information

Ca release is on track. It is targeted the 21^st of March. The TPCE master branch migration to Ca is in good shape.
Registration and topic submission period is still open for the next DTF (Developer and Testing Forum) which will happen in San Jose the 2^nd and 3^rd of May.

Code Review

Changes merged since last meeting on transportpce-models:

“Add a set of OpenConfig yang models” was backported to stable/potassium branch to allow lightynode building. It has been merged during the meeting.

Changes merged since last meeting on transportpce:

“Switch all func tests to lightyNode” & “Add a tox profile to install Lightynode”: now, all TPCE functional tests are fully migrated to lightynode, netconf simulator based on lighty-netconf-simulator (PANTHEONtech/lighty-netconf-simulator: lighty.io NETCONF device simulators and libraries (github.com)).
The relation chain dealing with the refactoring of the rollback process in the renderer

Changes transportpce under review:

mainly the relation chain related to the Ca migration
- “Make lighty non-voting in CI”: Because the migration of lighty core is still under process, we need to disactivate lighty voting, temporarily until lighty is migrated to Ca. The lighty build issue (availability of RpcService) has been fixed during the week. As soon as Lighty is ready for Ca, we are able to revert this change to reactivate TPCE lighty build.
- “Bump upstream dependencies to Ca” is now ready to be merged. The merge of Smartopics' changes has implied to rebase this change which took times.
- “Fix Renderer Junit tests exceptions": during the execution of the renderer Junit tests, lots of exceptions were raised. Gilles repaired and adapted the Junit tests so that they can run without any exception. This also allowed to rebase Jonas change “Only verify PreFEC BER if end node is Xponder” which now passes the gate.
- Guillaume is working on different changes to apply code style rules : “Refactor common CrossConnectImpl”, “Fix inventory listener code indentation” & “clean up nbiNotifications after Ca Bump”
“Refactor Tapi 2.4 final touch” and “…. final touch2 ” shall be squashed and then will be ready to merge.

Other

An update of upstream dependencies for Ca GA, is currently ongoing. It is needed for the migration to Ca.
Everyone shall try to update his Jira Ticket so that we could close the Sprint after the migration to Ca has been accomplished (Please try to update before next week meeting).
Nathan mentions that they still experience some issue with Netconf session on devices from TransportPCE, when at the same time other sessions are in place to catch PMs on devices; whereas the device is supposed to support multiple (up to 18) sessions at the same time.

2024-03-07

General information

Topic submission period is still open for the next DTF (Developer and Testing Forum) which will happen in San Jose in April.
Ca preparation is ongoing with the target of 21^st of March

Code Review

Changes merged since last meeting on transportpce-models:

“Bump upstream dependencies to Ca”

Changes merged since last meeting on transportpce:

“Add SpectrumFillingRule check in postAlgoValidator “, “Refactor PCE OperatorConstraints “ and “Reindent PceTestData Junit”
“ Allow the client to specify ROADM service SRG port “ & “Address yang name change for transportpce-pce” changes from Smartoptics
Also Christophe's change on "Complement sims configuration files"

Changes transportpce under review:

“ Bump Upstream dependencies to Ca“: the migration is now OK and this change passed the gate after Robert solved a bug related to RPCs handling in the Netconf project. This change will be merged very soon to make sure that the transportPCE is correctly migrated to Ca. The developer and user guides will need to be upgraded to reflect some of the changes that were made in the way to connect node with Netconf. Java interface related to yang modules describing RPCs do not exist anymore. Gilles moved all RPCs in specific classes, having one RPC implementation per Class. Thus, can RPC be registered individually. Gilles had to rework a lot of Junit tests. Implementation of Listener and DataTreeChange implementation is also modified. With current version of lighty core, it is not possible anymore to build lighty using Ca dependencies, due to the RPC implementation change that is associated to ODL (need mdsal.binding.api.RpcService). One shall also note that currently Openapi does not work anymore in Ca. This issue shall be solved in R 7.0.2 of Netconf.
“ Only verify preFEC BER if end node is Xponder“ change from Jonas makes a lot of Junit test failing. Gilles is currently working on this, rebasing this change on the migration to Ca to benefit from the Junit tests he has refactored.
Joakim changes are currently ready, but we will merge the Ca migration first, and rebase them after.

2024-02-29

General information

Topic submission period is open for the next DTF (Developer and Testing Forum) which will happen in San Jose in April.

Code Review

Changes merged since last meeting on transportpce-models:

OpenRoadm 13.1 models and the set of OpenConfig models have been merged.

Changes merged since last meeting on transportpce:

“Add Customer name to pathComputationRequestInput“
Guillaume changes on Tapi refactoring
Jonas changes were merged (“Allow 0 dB spanloss in power management” and “Fix persistence when karaf is restarted”).
“Determine preferred node/port from PCRI“, “New Package dealing with Rollback process” and “Refactoring rollback when service create fails”. Gilles hopes this will not have too many side effects on his change related to the migration to Ca since this is really complex. "Bumb upstream dependencies to Ca" is now in merge conflict state...

Changes transportpce under review:

“Refactor PCE operator constraint “ has been proposed by Guillaume to optimize for loops.
“Address yang name change for transportpce-pce “
“Add a tox profile to install lightynode“ , “Switch all func tests to lightynode” and “ Complement sims configuration files” have been proposed by Christophe in order to migrate the CI/CD chain towards Lighty.io. => ready to be merged.
“Refacto to Tapi2.4 Final touch” correspond to latest work associated with the migration to Tapi 2.4 , introducing spectral information in Neps. This change is not for review and has been set as WIP. It has been pushed for backup.
“Only verify PreFec BER if end node is Xponder” currently fails. It will need to be rebased on Gilles's change on the migration because most of JUnit tests have been redone.

2024-02-22

General information

The annual LFN 2024 mentorship program has been launched. Selection progress is already started. The deadline for submission is the 17^th of March. This program allows mentoring students that are willing to contribute to a project.

Code Review

Changes merged since last meeting

“Fix warning message when service creation fails” & “Fix ConvertORToTapiTopology Upper/Lower Freq bug”

Changes transportpce-models under review:

Change associated OpenROADM device13.1 models and OpenConfig models could be merged

Changes transportpce under review:

4 changes associated with TAPI refactoring are still in review status. 3 of them could be merged. “Outsource TAPI test constants from test utils” may not be merged according to the comments that were made.
“Fix OLM rollback” and “Allow 0 dB spanloss in power management” could be merged. For other SmartOptics changes, we may wait since they do not pass the gate.
“Fix persistence when karaf is restarted” : Guillaume had a look at it and asked for an adaptation. He suggested to adjust the script used at initialization.
Smartoptics provided 2 changes to correct RollBack function : “New package dealing with device Rollback” & “Refactoring Rollback when service create fails”. Bala did not have time to review it.
SmartOptics proposed an additional relation chain including “Determine preferred node/port from PCRI” and “Allow the client to specify ROADM service SRG port” which address Bala’s bug TRNSPRTPCE-176

Other

Christophe and Gilles succeeded in integrating OpenConfig Models and the xml configurations in Lighty Node. They still have an issue to build the right xsl transform to expose the configuration in the Operational DataStore.
Gilles confirms that we can change in OpenROADM MSA latest models the import of of iana-safi to iana-routing-types version 2022-08-19 which is the latest.

2024-02-15

General information

Next DTF Development and testing Forum will be hold in San Jose, collocated with ONE summit the 2^nd and 3^rd of May

Code Review

Changes merged since last meeting

FixConvertORToTapiTopology Upper/lower Freq: bug detected during the code refactoring step.

Changes transportpce-models under review:

“Add OpenROADM device 13.1.0 models” could be merged
“Update the iana-afn-safi import in lldp module”: could also be merged, and no objection were formulated during the meeting.
“Bump upstream dependencies to Ca” was pushed by Gilles this week to start the migration to Ca, and no specific issues were identified on this side.

Changes transportpce under review:

Smartoptics provided 2 changes to correct RollBack function : “New package dealing with device Rollback” & “Refactoring Rollback when service create fails”. Bala has been asked to review. The new functionality may be easier to test on devices than simulator. Gilles confirms that we can wait before merging it. In this change, multiple interfaces/classes were proposed to handle the rollback process. It is not that easy to figure out whether the new rewritten callback function will work, and the best option to test these changes is to test them on real equipment.
“Fix OLM rollback” could be merged now. Not really dependent on the previous relation chain
“False warning message when service creation fails” could be merged.
“Only verify PreFEC BER if end node is Xponder” pushed by Jonas avoids reading PreFEC BER from end nodes when creating a ROADM-to-ROADM. Need to be check because one Junit test fails... Issue on the gate?
“Add SpectrumFillingRule check in postAlgoValidator” : Olivier shall cleanup the Junit test before we merge it.
Gilles initiated the migration to Ca with “Bump upstream dependencies to Ca”.
“Add a Tox profile to install lightynode”, “Switch all func tests to lightynode”, “Upgrade LightyNode sim version” are still under review.
“complement sims configuration files” passed the gate but seems to fail right now with the same issue that we experience with some other changes. Christophe continues debugging the OpenConfig xml configuration files. He experiences some issues with some augmentations, especially for components, and more specifically psu.

2024-02-08

General information

ODL distribution for K-SR2 is released. Release notes have been updated.
Gilles started the migration of master branch to Ca (targeted mid-March) which seems to have important side-effects on TransportPCE current implementation. Hope it will not be too much difficult to adapt...

Code Review

Changes merged since last meeting

Olivier's relation chain on tapi 2.4 models migration. TPCE is now implementing T-API 2.4 models.

Changes under review:

Guillaume has initiated a refactoring activity on TAPI code:
- “Refactor Tapi Connectivity Utils” which simplifies the code style has been reviewed by Gilles,
- “RefactorTapi ConvertORTopoToTapiTopo” has been reviewed by Gilles,
- “RefactorTapi ConvertORToTapiTopology” : Gilles has reviewed it, Olivier started reviewing it but has not finished,
- “Outsource TAPI test constants from test utils” has been reviewed by Gilles who provided some comments,
- “Refactor Tapi ConvertORTopoToTapiTopo junit” & “Refactoring some Tapi Junit tests code ” have been reviewed by Gilles and merged in session
Smartoptics provided 2 changes to correct RollBack function: “New package dealing with device Rollback” & “Refactoring Rollback when service create fails”. Bala has been asked to review. The new functionality may be easier to test on devices than simulator.
New changes from Olivier related to the wavelength selection issue raised by NTT: “Add Customer name to path Computation”, and “Add SpectrumFillingRule check in postAlgoValidator” are 2 changes to handle a service creation according to a specific wavelength. As wavelength can not be provided as a constraint in service-create, we propose to leverage the spectrum-filling feature of Controller-behaviour-settings which was initially proposed to partition the spectrum across customers. Thus the wavelength constraint can be expressed as a specific customer profile (using customer-name in service-create) for which we dedicate a specific spectrum as specified in spectrum-filling-rules.
Functional tests migration to lightynode:
- “Add a Tox profile to install lightynode”: Gilles pushed this change to allow installing lightynode according to any version (specific release or specific snapshot binaries)
- “Switch all func tests to lightynode”: Christophe's change allows to switch from Honeynode to Lighty for all the tests (1.2.1, 2.2.1, and 7.1). In tox.ini , it allows adopting lightynode sims for remaining 7.1, hybrid and with_docker tests. It also adapts tests using netconf connector instead of restconf one as soon as Lightynode does not support restconf api
- “Upgrade LightyNode sim version” upgrades Lightynode version to manage notifications for 1.2.1 and 7.1.0. Now integrated inside the previous one. Will be abandonned.
- “Complement sims configuration files” is complementing configuration files for 7.1 configuration (pm description and pm list)

2024-02-01

General information

Potassium SR2 has just been approved by the TSC and the ODL distribution shall be released soon. For TransportPCE, the last changes that were pending (release note update in docs project) have been merged by Guillaume.
Gilles will probably start the migration of Master branches to Ca next week.
Robert informed the community about a migration to Java 21^st planned in Ca SR1, and imposed in Scandium.

Code Review

Changes merged since last meeting

Changes related to the K-SR2 release and backported from stable/potassium branch
Matej's change that was pending for a while (“Don’t use notificationListener (NodeRegistration)”)
Removal of transportPCE swagger feature that was an old workaround for developer that is not needed anymore.
Synchronisation of master branch of transportpce-modes to K-SR2

Changes under review:

Olivier's relation chain on tapi 2.4 model migration
- The relation has been reworked end of last week. Gilles reviewed it, and Olivier has addresed most important comments made by Gilles, and then reabsed again the chain. Gilles is about to merge them if nobody has any objection.
Tapi code refactoring: Guillaume has initiated a refactoring activity with several changes:
- “Refactor Tapi Connectivity Utils” which simplifies the code style,
- “Outsource TAPI tests constants from tests utils” and “Refactoring some Tapi Junit tests ” which simplifies the structure of the tests
Gilles pushed several changes on transprtpce-models
- “Add OpenROADM device 13.1.0 models” allows to have 13.1 OpenROADM device models for future developments in tpce.
- “Update the iana-afn-safi import in lldp module” is solving the issue of relying on an obsolete model. Long discussion about possible impact of having a change of namespace with the iana-routing-types model...
- “Add a set OpenConfig yang models” is adding a set of OpenConfig models to tpce, which correspond to what we see in some pre R1.0 equipment. It is still under review.

General topic

Olivier has provided a draft of the OpenConfig device configuration for Lighty, which consistency shall be tested.
reminder: url to use to have swagger is http://<server>:8181/openapi/explorer/index.html

2024-01-25

General information

Gilles released TransportPCE artifacts for K-SR2 + prepare its integration inside ODL distribtuion + prepare the release note on docs project

Code Review

Changes merged since last meeting

New relation chain from Gilles on stable/potassium related to the K-SR2 release preparation

Changes under review:

Same Gilles's K-SR2 related relation chain backported from stable/potassium branch to master
Olivier's work on tapi 2.4 migration
- Olivier shall rebase the T-API relation chain after Gilles's changes are merged
- Guillaume considers that the relation chain can be merged. Gilles will have a look at it before.
Addition of a set of openconfig yang models in transportpce-models repo which corresponds to what we see in some pre R1.0 equipment.

General topic

Olivier is working on correcting and completing the OpenConfig device configuration for the lightynode simulator, making it closer to what we observe with equipment in the labs.

2024-01-18

Code Review

No change merged since last meeting

transportpce changes under review:

“Don’t use notification Listener (Node registration)”: The missing Netconf features will be added in AR-SR4, K-SR2 and Ca-GA. Could be merged after the dependency bump
Status on Olivier's work on tapi 2.4 migration
- Olivier addressed Guillaume's comments in the relation chain associated with migration and refactoring. Some of the tests do not pass. It will be corrected according to the results of the gate that will help identify where (at which level of the chain) the bug was introduced

General topic

Sprint 34 was initiated. All non-closed issues of previous sprint have been included. An additional issue #774 was introduced : Create an OpenConfig configuration file. Some JIRA ticket describing OpenConfig models to be introduced in tpce models need also to be created.
Christophe made a small project to illustrate the bug we have in lightynode for openroadm-device 7.1 models (handling when statement). He reported the issue on the odl-kernel list. The Issue is identified in the yangtools project, but it seems difficult to solve it, and no solution will be provided in a short time. 2 workarounds have been proposed.
Bala is trying to fix the way we are building A and Z end containers in service-create and temp-service-create so that he can retrieve the operational mode supported by external pluggables to address IPoWDM case.
He also started to implement the power control RPC to address IPoWDM use case.
Bug: When we specify the port to use in the service-create (for SRGs ending an optical tunnel), tpce is currently picking the port based on first port available, not considering the port specified in the service-create. Gilles started to work on this bug but did not have the time to work on any use case. He asked Bala to create a JIRA ticket to better understand the associated use case.

2024-01-11

Code Review

changes merged since last meeting:

3 changes related to the bug TRNSPRTPCE-772 reported by Joakim
- “Spectrum assignment skipped the last node in path ” proposed by Joakim
- “PCE graph spectrum assignment slight refactoring”: small refactoring pushed by Guillaume
- “Add new functional PCE test to check bug 722”: new small pce functional test suite proposed by Gilles to check the effective correction of the bug.

transportpce changes under review:

Status on Olivier's work on tapi 2.4 migration
- It's a bit of a mess in its relation chain: Gilles will help to realign correctly the different changes.
- Still in work in progress. Need to address Guillaume's comments on few changes

General topic

Bala mentions that we may need to implement OpenROADM 13.1 device models in TransportPCE. Gilles mentions that it is not an issue to add device models 13.1 in transportpce-models repo. He is also in favor of deprecating the oldest ones (1.2.1) which are not used except in TransportPCE functional tests. Gilles will communicate the intention to cancel the support for R 1.2.1 models on the dev list.