2021 meeting minutes
Meetings connection details
available at TransportPCE meetings
https://wiki.opendaylight.org/display/ODL/TransportPCE+meetings
12/16/2021
Code Review
Since the last meeting, a large number of changes have been merged:
Changes associated with OTN implementation with intermediate switches
Migration of some of the functional tests (topology) towards RFC 8040
Changes associated with TAPI implementation: the sequence of the changes has been changed. The initial change associated with OTN link removal has been modified so that OTN links are not removed from the Abstacted T0 Multi-layer topology.
Reactivation of Unitary test for renderer.
“Remove forced revision imports in few tpce models”
”Unofficial regenerator capability support in Yang” : some identities supported by some suppliers are no more the official one (R7.1). We agreed with Bala to modify the official model but some details need to be provided in a notice so that users understand it. Guillaume made what was needed to close the change which was merged.
_Power control support for 87.5 GHz spectrum width: Gilles reworked it, rebasing it, and solving issues appeared after the merge of the renderer unitary tests activation.
“Increase transportPCE tox verify job timeout” allows solving some issues with the gate
Guillaume opened a comment on the already merged change: ”Tapi Link creation refactoring”: Guillaume solved an issue associated with the support to lighty.io in another change. So, the comment may be closed now.
All these changes have been backported on the stable/phosphorus branch.
The following changes are still under review:
TAPI changes concerning notifications:
a new change “Fix few NPE in TAPI implementation” has just been pushed by Christophe and Gilles to fix some NPE in TAPI implementation, when creating 100GE service over TPDR
Javier rebase the TAPI series related to TAPI notifications
Reactivation of lighty : “Add and fix Phosphorus lighty.io”; A few things have been changed in the PortMapping. Need to be reviewed again.
"Enforce Pylint in CI via tox is mergeable". The Pylint configuration has been customized to make it fit with our needs. Fix me has been disabled on redundant code. It may however need to be rebased.
“Use environment variable for OLM timers”: this change proposed by Gilles and Christophe will allow to set specific OLM timer values according to specific environment variable set or not before launching Karaf. It allows decoupling the build step from the runtime step. Thus, we can change OLM timer values without having to recompile again, depending on the usage (test on real equipment versus test on HoneyNode simulators.
Other Open changes in WIP:
Migration to Service model 10.0: we are waiting until R10.1 of OpenROADM is publically released to migrate directly to this last that brings some corrections that will ease the migration.
Final step of the migration of functional tests towards RFC 8040 has been put on hold
Q&A
Next DDF: preliminary planning has been sent by Guillaume, who is proud to have 5 subjects proposed around transportPCE.
It was the last meeting of the Year 2021. Next TransportPCE meeting will be hold the Jan 6, 2022 . In the meantime, Happy New Year's Eve
12/09/2021
General information
P SR1 : approved and available for download. P SR2 is planned for the end of January.
A question was sent by Santh about a compilation failure issue with TransportPCE. Guillaume answered, considering it was probably a Tail-f issue. The issue has been partially solved, and the device is now correctly mounted. However there are some remaining issues. Guillaume identified that it could be an issue with NETCONF (Santh uses Al distribution). His suggestion is to migrate to a more recent version.
Code Review
Since the last meeting, only one change has been merged: “Fix remaining Pylint warnings apart fixme & dup”
There are more than 40 open changes which are in review.
Gilles and Christophe have agreed on a strategy for tomorrows demo. The relation chain of changes related with T-API will be split into 2. They have reordered some of the changes so that they can merge earlier the 3 following changes : “Adaptation of xpdr mapping from OR to TAPI”, “Tapi link creation refactoring” and “Fix tapi get-topology-details functional tests”. Remaining chain of OTN changes has been put on top of this and is in the last review step.
We are currently experiencing some issues of the gate.
The fix proposed by Javier to solve the issue associated with test 16 seems not to work. We still experience the issue. Javier is still trying to simulate the issue locally.
Some other changes fail due to either an issue to launch sims121 or an issue launching TransportPCE controller for tests221: according to Guillaume, this could be due to resource starvation issue (at the hyperviser level). He suggests we raise a ticket.
Christophe highlight the fact that any change submitted should include some functional tests as soon as a new functionality is introduced. This shall be applied to the case of the change that were initiated by Peter. Javier agrees with this and will work on it.
“Reintroduce nb-bierman02 support in new functional tests”, “Migrate topology funct tests to RFC 8040 step 1, 2 & 3” have been proposed by Guillaume and are under review.
Migration of Service Model to 10.0: Gilles proposed 3 changes on this topic:
“Remove fix revision imports in tpce yang models” prepares the migration by removing revision forcing in TransportPCE yang models, and will allow to always retrieve the latest model revision available
“Upgrade to service Model 10.0” is the model migration in itself. The issue encountered compiling the new service model has been raised in OpenROADM MSA (Issue 853). PR 855 is solving this. It was presented this afternoon, and nobody complained. So it shall be part of service model 10.1. Olivier suggests to put this change in standby and wait until we get service model 10.1 available. This change breaks the API so that it will proposed for Sulfur.
“Remove service-endpoint grouping” does basically what we requested in OpenROADM Service model 10.1 through PR855. As observed on gerrit, it solves the compiling issue.
Guillaume proposes to rework the change proposed by Bala “Model 7.1 change to support regenerator capability” so that we can have an “acceptable” solution
“Power control support for 87.5 GHz spectrum width” : Shweta will try to address this changes before going for holidays.
Q&A
Next DDF: 5 subjects have been proposed around transportPCE by Nokia and Orange.
Last public TransportPCE meeting of the year next week on the 16th.
12/02/2021
General information
P SR1: Christophe successfully released transportPCE in P SR1 the 30th of November.
The main features introduced in SR1 correspond to around 50 changes (100GE over OTN switches, T-API improvements, start of migration of functional tests toward RFC 8040, and a number of fixes).
P SR1 was approved.
Gilles updated docs project with the TransportPCE release notes.
P SR2 is scheduled for the 27th of January 2022.
RESTCONF api will change. Guillaume has initiated some work to migrate from Bierman to RFC8040 which will help to support this migration.
Code Review
Since the last meeting, the following changes have been merged:
“Fix some/more Pylint warnings” and “bump network dependency to 2.0.11” solves the issue of the Hexadecimal trace in the Logs.
Gilles will backport these changes on stable/phosphorus branch
change 98757: “Set xponder-type in the mapping for transponders”:
for Xponders we put the type of xponder (mux or switch) which is embedded in the list of xponder in the device configuration, with the exception of transpornders, we are still not putting anything in the mapping since in Release 1.21.1 this list and types were not provided by the devices. So when we don’t have the list of transponder we will put in the portmapping the “tpdr” type.
change “Fix tapi get topology details functional test”: the change instead of using Databroker use network transaction service which allows providing a more reliable behavior
change “Tapil ink creation refactoring”: provides a unification of tapicreatelink and TapiLinkUtil and avoids code duplications
change “Adaptation of XPDR mapping from OR to TAPI”: solves some bugs and provides some modification of the way NEP are created
change “OTN link deletion from T-API topology”: transforms OTN link to TAPI connections to be in line with TAPI implementation
change “Unification iNodeEdgePoint _N and iNodeEdgePoint_N”: allows to use the same naming convention for this NEP in both the T0 and the T0 Full Multilayer topology.
change “OpenROADM Xpdr link TAPI listener”: allows creating links directly from the update in the OpenROADM topology rather than using specific rpc to create roadm to roadm links and Transponder to roadm links
Christophe raised the question about the deletion of T-API links. => we need to confirm if the rpc delete link runs correctly for openroadm topologies
change “Tapi connectivity upgrade”: Implementation of 100GE services in Tapi as an Ethernet service and OTN services for Muxponders and switchponders
Javier has also been working on the TAPI notification services already addressed in the previous meeting.
OTN switch changes series: Gilles will start addressing the comments received so far very soon.
Guillaume also pushed 3 changes on Pylint and functional test migration towards RFC8040. “Enforce Pylint in CI via Tox”, “Fix remaining pylint warnings apart fixme & dup”, “Migrate topology functional tests to RFC 8040”
Gilles started the work to migrate the service models to R10.0. But we have an issue and Christophe raised the corresponding issue in OpenROADM community.
Q&A
Next DDF:
Gilles and Christophe proposed as a topic, an Overview of transportPCE feature, including a short demo
Nokia will also propose 2 topics: one about what they have done in transportPCE, one about what they plan to do next
Guillaume has proposed one topic on release management, and one about tox jobs parallelization.
11/25/2021
General information
P SR1: the release has been postponed two weeks later (MRI project not ready yet) to the 2nd of December. transportPCE shall then be staged between the 2nd and the 9th of December
Releng project shall move to Maven 3.8.
Transportpce mailing list: a mail was sent by Shanth with regards to issues experienced when connecting to Cisco devices. Some additional logs were sent but logs are only partial. Guillaume shall provide an answer.
Last TSC meeting summary provides a link to connect to Slack dedicated to discussions with the ODL community.
Code Review
Since the last meeting, the following changes have been merged:
the beginning of the otn suite (merged and backported to stable P branch)
the bump to netconf dependencies to 2.09, and then to 2.0.10. As a result, now part of TransportPCE LOG includes Hexadecimal codes when connecting to a NETCONF device. This issue has been raised to Robert who shall fix it very soon.
Guillaume made some changes related with migration towards RFC8040, “Migrate portmapping functional tests to RFC8040” (merged, to be backported to P), “Run pre-commit autoupdate”, “Enforce Autopep8”, “Remove u prefix obsoleted by python 3”, “Fix some Pylint profile” (could be merged), “Enforce Pylint with a score >=8.97 in CI” (not to be merged)
”Fix null Supported CEP Qual. on TAPI node creation“ has still one comment to be addressed. A Jira ticket could be created to keep trace, allowing to merge this change
A lot of changes failed last week, probably due to instability with T-API tests.
“Fix tapi get-topology-details functional test” should solve this issue
All Changes related to the use of intermedaite otn-switches shall be rebased on this fix to check on the gate its positive effects.
T-API changes suite:
Tapi link creation refactoring tries to reduce code duplication, and includes latest changes associated with proposed link naming convention
Update TAPI connectivity service on port change : Christophe is currently testing it.
Gilles will backport all recently merged changes on the stable P branch.
Robert recently released Si SR3.4 to address some Netconf issues (Si SR4 is not planned)
We may experience an issue with Si SR3, because the test suite for tox parallelization implies to be activated on releng project, what has not been done yet...
Q&A
Next DDF:
we can submit some topics until the 4th of December here: https://wiki.lfnetworking.org/display/LN/2022+LFN+Developer+Event+Topics+January
Guillaume will propose a talk on the different way to manage projects (MSI/ MRI/self-managed) and another talk about tox test parallelization
Highstreet will probably propose a topic on ONAP and Tpce
Guillaume suggest to proposed talk to introduce the tpce roadmap
Javier would like to propose a talk about T-API development performed on tpce with probably a small demo and another talk could be about what Nokia plans for next transportPCE contributions.
11/18/2021
General information
TransportPCE was released on the 5th of November for Si SR3. Since we release TransportPCE (SM project depending only from MRI upstream projects) before MSI projects, it led to exchange on the community, and Robert highlighted that tpce could become an MRI project. We are not so clear on the different commitments that are associated with MRI project. Gilles suggests everyone has a look to the related documentation, and we'll discuss this point again next week.
https://docs.opendaylight.org/en/stable-phosphorus/release-process/managed-release.html
OpenDaylight distribution, integrating TransportPCE has been released on the 11th of November in the Si SR3.
Gilles proposed a change in the doc project to update the release note from the Jira tickets associated with Si branch.
Gilles is not sure that the parallelization of test is effective on stable/silicon branch. According to a mail from Guillaume, we may have to update Releng to make this effective. Gilles will discuss this with Guillaume.
P SR1: Gilles has backported some change from the master branch to P. However, waiting for the unlock of our stable/phosphorus branch before merging all associated changes.
According to the ODL schedule (Phosphorus SR1 Release Checklist), the branch shall be unlocked tomorrow. Christophe propose to handle P release. (Official distribution with tpce due on the 22nd of November)
An issue was raised by Shanth about the NETCONF connection between TransportPCE and their Cisco equipment. Guillaume shall provide an answer.
Code Review
Since our last meeting, almost 20 changes have been merged (T-API related, Tox/functional tests related, SONAR portMapping issues, as well as some changes from the OTN suite).
Change 97868 (Fix null Supported CEP Qual. on TAPI node creation) was merged while one comment from Guillaume had not been addressed. It was decided to open a new Jira Ticket to track the comment which could be long to manage.
Change 98531 (Bump netconf dependency to 2.0.9) could be merged.
Status on the OTN switch series:
Guillaume already merged 3 of the changes. Gilles has reworked all changes considering the comments made by Guillaume and Christophe.
Apart from ”New functional test with intermediate otn switch” for which a test needs to be renamed from 12 to 14, the other changes can be merged as they are.
T-API change suite:
Gilles and Christophe started to test « Update TAPI connectivity service on port Change ».
Javier will send his postman collection associated with T-API so that Christophe and Gilles can test it.
In the update of the connectivity-services when a state changes, all services supported by the out of service connectivity service shall be updated. We agree on this expected behavior but it seems that right now, when a photonic layer service is out of service, the upper layer services’ state is not refreshed. Javier will check this.
In “Migrate portmapping functional test to RFC8040”, Guillaume changed the url for tox tests. Gilles would prefer to wait until he has merged some of the changes of the OTN suite, prior to merge this one to limit the amount of work.
Change 98001 (Reactive renderer UT): Gilles would prefer waiting that all series on otn switch would be merged before rebasing this change, to minimize conflits to manage.
Change 97599 (Model 7.1 changes to support regenerator capabilities): Bala needs to elaborate the Jira Ticket before we discuss if something more needs to be made.
Q&A
We could not demonstrate 2x200G in Supercomputing (no available product at the time of the demonstration). A video was displayed.
OFC 2022 demonstration : Bala discussed with Andrea. Regen support on devices might shift from January which was first availability to a later date. 3x100G might be supported by some equipment at the time of OFC demonstration. UTD has still in mind to demonstrate path computation relying on GNPy
DDF will be beginning of January (https://wiki.lfnetworking.org/pages/viewpage.action?pageId=53609047). Subject to be presented must be submitted before the 3rd of December. Any proposal is welcome.
11/11/2021
November 11 being a public holiday in France, this TransportPCE public meeting is cancelled.
10/28/2021
General information
ODL Phosphorus Release: Guillaume, as the TSC chair, made a quote for the marketing message. TransportPCE is mentioned.
Our Jenkins/Gerrit and JIRA may be unvailable next week due to maintenance operations. Beware, since we must release Si SR3.
Si SR3: The release date is planned for the 4th of November. The release of Kernel project is planned next Monday. Then the release of self-managed project is planned for the 4th of November. Our stable branch shall be unlocked on Monday. The time slot we have is between the 1st and the 4th of November. Shweta shall cherry-pick today the changes on stable Si branch. She identified 20+ changes to be backported. TransportPCE Project version need to up increased: for Si SR3, it shall be 3.3.0-snapshot (Phosphorus is 4 and S is 5).
Gilles, Christophe and Guillaume being on PTO next week, Shweta will take care of releasing TransportPCE for Si SR3.P SR1 is planned for the 22nd of November. Gilles suggest that we regularly (every weeks) backport changes done on master branch to P. Gilles already made this. Currently, most of changes are failing due to a bad project version in one change related to swagger feature.
Orange made a demo of HoneyNode to UTD. UTD proposes to make some development around PM generation in order to have an optical node emulator a litle bit more dynamic.
Code Review
“Initial OR T-API mapping : topology” is not a new change merged this week, but supports some ongoing discussions on the way to compute the SIP UUID. At the end we decided to follow Javiers approach on the SIP naming convention
“Refactor KAraf builds for test and tox.Ini” and “Remove upstream dependencies version ordering” have been merged
Extend T-API feature Documentation needs to be rebased on top of master. Then it can be merged
Update developer guide for Si and P changes : Olivier will add a comment on the creation of ODUC-4 (comparison with OTUC-4)
Fix bug in OLM spanloss computation needs to be reviewed before being merged. Christophe proposed to review it.
Reactive renderer UT : creates a lot of merge conflict. Gilles will work on it after he addresses the OTN serie. It will be put above the otn serie.
T-API changes chain : all the changes below the T-API functional test, apart from the Junit tests change, solve bugs. They are ready to be merged. All the changes above need to be reviewed. Javier is ready to demonstrate their behavior during next meeting. Christophe mentions that we may need to do some refactoring to avoid code duplication. This could be made in parallel to code reviews.
Another change is not in this relation : “Fix rule format for service interface Point”. It has been dissociated to simplify its merge.
OTN switch changes chain : Gilles will move to the bottom the changes associated with bug solving to reduce the relation chain and simplify its merge.
Q&A
Gilles confirms that ODUC4 and OTUC4 could be created at the same time. We did not identify use cases where OTUC4 and ODUC4 could have different end points. If TransportPCE had to be modified, it would not be so trivial
10/21/2021
General information
_Phosphorus is officially released. We still have a couple of issues with the document but it can be published Later on.
_We will have Si code freeze for SR3 next week.
_Gilles has been successfully elected as the new PTL
Code Review
_Gilles pushed a change that removes upstream dependencies version that were overriding versions managed in ODL parent. This is needed notably for Mockito. This change could also be cherry-picked in Si
_He has pushed another change to solve a bug raised by Jonas and addressing User story 172 : when not having PM for OTS interfaces in the device, get PM method would generate a NPE.
_Javier pushed a number of changes that bring some fixes to some bug we identified implementing MDONs use case. Guillaume is fine with them but wants to wait until Gilles or Christophe
_Phosphorus t-api documentation : Javier added the description of the extensions he made for t-api module. This change shall be put at the bottom of the stack so that it can be merged before we merge the other changes associated with fixes.
_Olivier also pushed a change associated with the documentation to include both Si and Phosphorus changes.
_Gilles made a change to reactivate unitary test for the Renderer. It could be cherry-picked on Si. This change shall not be merged before we merge all the change suite (23 changes) associated with end to end handling of high rates OTN services. These 23 changes have passed the gate. Gilles will try to address all comments at the same time to avoid overloading the gate.
Gilles has observed that some functional test (the one used for T-API) are randomly failing.
We currently experience troubles with the infrastructure
10/14/2021
General information
Phosphorus : Gilles took care of releasing P. It will be officially announced and we just need to provide the documentation by the 18th of October.
Javier will work on the T-API features description, Orange on the update of the documentation to mention that we do not anymore rely on intermediate APIs to create/delete services
Si SR3 is in preparation. The last change addressing the disconnection of nodes will need to be backported as well as some refactoring provided by Guillaume
Shweta volunteered to do the work to release Phosphorus.
Guillaume proposes to roll the role of PTL, and to engage a new PTL election. This will be launched either through a vote in Confluence, or through a mail thread.
Code Review
_Javier raised a JIRA ticket MDSAL-694, as he discovered some issues in importing some TAPI models in the MD-SAL which is related with MD-SAL binding (according to Robert).
_Javier provided some changes to correct some issues with T-API :
· Code refactoring for OR-TAPI mapping (Node NEPs and SIPs states , cleaning)
· Functional tests for new TAPI module (Junit test)
· Fix wrong CEP protocol qualifier (bug fixing)
Last three are based on Peter contribution
· Update topology Service on port change
· Update TAPI connectivity service on port change
· Send TAPI notification to Kafka
_Javier also started to update TAPI version from 2.2.1 to 2.2.3 but did not push it.
_Christophe has started working on “TAPI functional tests”, a functional test suite to test TAPI connectivity service. He noticed in some cases some strange behavior on the topology : sometimes it is not in line with OR topology. HE still needs to investigate to understand the reason for this.
_Guillaume solved the Swagger issue removing the OpenROADM OSPF models. He has finished the portMapping refactoring which was partially reviewed by Gilles.
_AT&T is still observing Cisco Mounting issue in its labs. The Bug is probably in Yang tools or MD-SAL. We are asked to raise a bug if we want this to be solved.
_Model 7.1 change to support regenerator capability : we agreed on keeping proposed solution (added identity in OR model) provided that we mention that the model deviates from original version and that we create a JIRA issue to use augmentation in the next step.
_Fix : disconnection of netconf node : Gilles provided a bug fix to address an old bug which has been raised by Robert. Ttrying to connect a node unsuccessfully, as the node is incorrectly connected, it does not appear in PortMapping and in the topology so that when it is disconnected, tpce tries to remove an non existing node (in topo and portmapping) which creates a NPE.
_Gilles has worked all the week to solve issues with functional tests on the last relation chain he pushed, addressing end to end OTN support in more complex configurations (Intermediate OTN switch on the path, and additional use cases ….). This was also an opportunity to add an additional test suite to test these new use cases.
10/7/2021
General information
Phosporus release was approved last Friday.
We were supposed to release by tomorrow. Guillaume and Gilles made the job, but we experienced a lot of failures with the gates : jobs start but never end.
Code Review
The changes regarding PortMapping refactoring have been merged (step1 & 2). We are now using constant in PortMapping messages.
Gilles has pushed long series of changes addressing end to end OTN services, using various end terminals (transponders, muxponders, switchponders).
Part of the relation chain is dedicated to 100GE over OTN switches, another focuses on managing intermediate OTN switch in the middle of path for low order services.
Swagger removal : the proposed change to reintroduce Swagger does not work. It seems Swagger was removed quite a long time ago : it is already not present in Silicon.
We need to investigate deeper on this subject : Swagger is really needed to debug, but the reason why the proposed change does not work is not that easy to understand.
Issues
Regen issue : Bala tried several workaround to solve the issue associated with the lack of identity for regen
Deviation is not supported for identities. Bala also tried to add a file providing needed identity using the same name-space as the file where the base identity is provided, but this does not work.
Guillaume propose to provide the missing identity through an augmentation in an additional file, and to force through a curl command the capabilities of the equipment.
We should also ask to the vendor that use this missing identity in their devices to use an augmentation.
At the end, as this generates a lot of work we decided that we can probably simply add this identity to the file. But we should update the description of the associated Jira ticket
Cisco models issue : Phosphorus bump brings an issue in the way some the tailf-ncs-plan@2019-11-28 yang model is imported (associated with a YangTool issues)
09/30/2021
Code Review
_Refactoring of the port Mapping relation chain has been merged. This includes “Add a portMapping Util class”, “Fix most PortMapping Sonar Issues”, Use constant for PortMapping log messages step1/2/3
_Guillaume is still working on other changes to reduce the cyclomatic complexity (PortMapping Refactoring changes)
_Christophe worked on B100G E2E functional test that Guillaume considers as ready to merge
_“Model 7.1 change to support regenerator capability” : deviation to model does not work. Another possibility would be to make an extension to the OR yang models, but Bala did not have the time to test it.
Gilles as created a Jira Ticket about the fact we have lost the swagger interface with transport PCE which is needed to use notably the API doc. Guillaume mentions that it use a lot of ports and might not be needed on every node in case of a deployment. AT&T confirms this is useful for them
09/23/2021
General information
Phosphorus is coming soon despite a few failures with BGPCEP project.
Code Review
7.1 models does not support the regen identity. OR community did not want to create a minor release just for this. It was agreed with the vendors to manually add this identity in 7.1 models (org-openroadm-port-types), to support regen feature.
If we don’t add this identity on the controller side, the controller will fail accessing the transponders.
Guillaume suggests we use deviation to add the needed enum, but we may generate some java object that is not exactly structured as the object resulting from manual add in the device models made by OEMs. Bala will try this approach.
For what concerns change 97600 rather than changing the models, removing the deprecated enumerations, we could probably use tox to patch the models and remove the deprecated identities.
Guillaume started working on “switch functional test to RFC 8040 url”. He investigated on solving portMapping SONAR issues (“Add a PortMapingUtil class”; “Fix most PortMapping issues”)
“Use constants for PortMapping log Message” (step1 & step2) also fix a few SONAR messages and bring some consistency in the way LOG message are handled.
Christophe worked on “Code consolidation for end to end B100G services” which addresses 400G transponder and 100G:400G muxponder use cases.
He has also provided on a test suite using 7.1 transponder versions and 2.2.1 ROADM versions : “B100G end to end functional tests”.
09/16/2021
General information
Phosphorus code freeze will come soon (still one issue in BGPCEP).
Guillaume proposes that we envisage rolling to take on PTL role and ask for potential candidatures
Code Review
_Gilles merged almost all the change associated with P migration
_Notification stream selection update is ready to go, now that conflict have been solved.
_Peter has pushed some code that shall be rebased on “Improve container support in tox.ini” which solves some docker issues
_“Fix portmapping sonar issues” has been pushed by Guillaume as WIP . He will address cyclomatic issues in a different change.
09/09/2021
Code Review
“Notification stream selection update” : Gilles managed to run the test manually, after making some adjustments, in some specific conditions. “Fix Hybrid functional test race conditions” change allows solving these issues and running the tests in a more stable way.
“Fix GNPy functional test” change made by Guillaume allows to fix the failing of some functional tests associated with GNPy
“Improve container support in tox.ini” : promotes podman rather than docker as the solution for container, because Docker ……
Christophe made a review on Phosphorus migration. Everything seemed to be OK, but since this morning he has observed some issues with Phosphorus distribution (associated with NETCONF and Karaf). Nobody on the bridge had to work with Phosphorus so that we don’t know if these issues affect others or if it is limited to Christophe’s environment.
Christophe also observed another issue with the management of the OLM timer which is now directly injected in the Blueprint. After changing the timer several time, the modifications are not considered. Guillaume mentions that he needs to restart the bundle, since blueprint modification are not take into account if this is not done.
Peter is experiencing some issues in pushing his changes, so that he will need more time to do so.
08/26/2021
General information
We had an infra failure on Jenkins which generated a global failure, but this is solved.
Si SR2 has been released.
Code Review
Guillaume has merged all the relation chain related to the parallelization of tox Test. In Jenkins, now, all tests are parallelized. A documentation is included (Change 97194). Locally, by default, tests are not parallelized.
“Refactor NMI notifications and ServiceHandlerImpls” & “Update NBINotifications” doc can be merged.
“Notification support for 7.1 devices” adds notification and tca listeners. This change has been there for a while and can be merged
“Notification Stream selection” has been updated, but Shweta needs some help. Modifying simulator’s xml configurations, changing Netconf stream from OpenROADM to Netconf does not solve the issue. Gilles will investigate.
“Power control support for 87.5 GHz spectrum”: Guillaume created a Jira ticket. Shweta will add some input to the renderer input container (nmc/mc)
“Use notification switching-pool in portmapping update” can be abandoned since we do not experience anymore the locking issue.
“Patch org-openroadm device 7.1” can also be abandoned since the serialization issue has been corrected in Phosphorus and Si SR2.
Guillaume made a change to fix unused dependencies warning now that we made a bump to phosphorus.
08/19/2021
General information
We are still waiting for LFN staff to release Si SR2. Dependencies have been changed. We will need to restage.
Code Review
_Update karaf cache nc-notifications YANG schema : we may be able to use an updated version of the yang models even if they are in draft mode.
_Notification support for 7.1 devices : some comments have been made and need to be checked.
_Refactor NBI notifications and ServiceHandlerImpl : Thierry has started addressing the comments and recommendations provided by Guillaume and Christophe. The terminology used shall be in line with the documentation (and reversely).
_Peter is still working on his changes : sending NorthBound notifications as the connectivity context is changing
_Guillaume has started writing the documentation associated with the test parallelization.
_OLM management : New comment from Jonas regarding the fact that we use different channel width for 200DPQPSK, 300G QAM8, and 400G QAM16. Jonas is right, as all this rates use the same Baud rate and have the same roll off factor (max 0.20), we should calculate the same width = 63.1*1.2 = 75.72 GHz.
08/12/2021
General information
Si SR2 is not ready yet for managed project. We can expect it to be released soon. Could be next week.
No news for Phosphorus (SR0 planned mid-September)
Bala had a discussion with Andrea and Janeth. They want to go for Supercomputing (End of November).
The base of the demo will be the same as the one presented in OFC 2021 with potentially new equipment (400G, 2x100G over a 200G, Regen).
We could probably focus on intermediate rate provisioning and on path computation for high rate relying on GNPy for the demo of both Supercomputing & OFC 2022
ODL marketing team (contacts @Casey Cain @Brandon Wick ) is also looking for demo to market : we could talk of the OFC demo (some elements have been publically released by UTD)
Some expectations also about 5G. OpenROADM is supposed to be a good candidate for 5G WDM backbones but this plan was not yet realized or tested yet.
http://www.gazettabyte.com/home/2018/3/28/will-white-boxes-predominate-in-telecom-networks.html
https://www.infinera.com/wp-content/uploads/HR-Operator-Strategies-for-5G-Transport-July-2020_WP.pdf
Code Review
Power control support for 87.5 Ghz Spectrum width : we need to clarify the constant we use (and probably adjust their names to make it more comprehensible) when we compare them with input.getwidth to make some decision and calculate the power for Flexgrid operation.
Fix OLM timer for ROADM crossconnect power setup can be merged
Guillaume has started refactoring the code of the OLM because it was raising a lot of Spotbugs issues and warnings (Refactor OLM PowerMgmtImpl class)
Thierry refactored NBI notifications and serviceHandlerImpl to address some comments and correct some issues. He has also updated the docs on NBINotifications
Notification Stream selection update : Shweta updated NetConfTopologyListener to read notification steams (OpenROADM or Netconf) supported by device and adjust the subscription accordingly.
07/08/2021
General information
Guillaume will be on vacation during 3 weeks. Christophe and Gilles will take their vacation in August.
Next week meeting is cancelled.
Phosphorus kernel project are almost ready, so that we could bump the dependencies. The code freeze which was planned beginning of August will be slightly delayed.
Si SR2 will also come at the beginning of August but shall be delayed.
Code Review
Guillaume has merged a few changes :
Remove usless Netconf dependencies from features
Remove odl-netconf-connector-all dep from feats
Change tapi topology to use correct states
He has also started working on the long relation chain we have to fix minor issues
OTN service delete – B100G, Update configuration files of 400G simus, Renderer Interface delete for B100G TPDR should be ready to merge.
A lot of change have been pushed for Test parallelization : Christophe mentions it is not easy to understand the approach followed, since it includes a lot of changes and the commit message are to succinct to understand the methodology. Christophe would have appreciated to have this work introduced. Guillaume propose to explain the approach during the demonstration.
2 changes from Thierry : Refactor NBI notification & add service listener tests, Add service listener to notify Kafka.
Thierry addressed some of the comments but still has to address important comments. This work shall be tagged as WIP.
Christophe pushed 2 changes:
_Modify spectrum assignment in PCE : refactoring of the way to manage the spectrum assignment using a specific yang models rather than specific Java classes
_Update path description with OpucnTribSlots (WIP) : We still have some issues to solve. Christophe tried to align the path description with what is needed by the renderer. A leaflist is used for the min and max tribslot. Bala agrees on the fact that this shall be split into 2 leaves rather than using a leaflsit. The modification will be made tomorrow morning.
06/24/2021
General information
Al SR4 has been officially released. We don’t have any news from Phosphorus middle point timestamp.
Code Review
Peter has pushed a change “Change Tapi topology to use correct states” to use correct operational and administrative state for ROADM to ROADM links, as well as Xponder to ROADM links. This code allows retrieving states from the mapping updates. Functions were added to get the correct state and insert it to the links. Some other functions are provided to convert OR states to T-API states.
7.1 Relation chain status :
Bala has proposed some changes to address comments on the changes of the bottom of the stack. Christophe has continued working on it to address all the chain. Guillaume has then moved the change about the 7.1 versus 71.0 version at the bottom of the stack, feeling it could be merged in a short time. The change hardcoding honeynode node type to avoid an issue with the PRBS test pattern “issue with maint-signal container” has been removed from the relation chain since it cannot be merged as it is.
Christophe thinks we can merge all the changes on the bottom of the task. Bala may still need to address a few comments for the changes on the top, which could be merged in a second step.
Change 96210 “Renderer interface delete for B100G TPDR” may include some dead code that we could remove. Shweta will update the code.
As Guillaume was the latest to work on the different changes because he moved the 7.1 change, he would like others to make the final review and merge the different changes of the relation chain.
Guillaume has started working on the test parallelization. This includes several changes (“Refactor tox.ini to prepare parallelized jobs”, “outsource PEC test scripts into a separate folder”, “Refactor tests launching procedure”, “Allow lighty build to listen to alternative ports”, “set different controller ports in tox test suites”, “Allow tox func tests suites to run in parallel”)
06/17/2021
General information
Peter Heringer has joined transportPCE dev team. He’s a student working with Nokia and will focus on T-API features.
AL SR4 : The branch is still locked. Christophe prepared a relation chain but we can’t push it.
OFC demonstration ran well, with a good level of attendance.
Developer forum : Gilles had the chance to have some exchanges with Robert about the tox job issues we are experiencing.
Code Review
Guillaume proposed some changes to automate trailing blanks handling and merged some changes associated with linters
Christophe commented on the fact that very long chains lead to complex issues as many contributors are involved, especially when some of the new changes proposed are not correctly rebased on latest ones. We should try to avoid this as far as possible.
Some modifications were pushed on “update PCE for OTUC4 ODU C4 and 100GE Mxpdr”. The associated chain is fine
Patch org-openroadm-device 7.1 yang model allows solving temporarily an issue with the MD-SAL, when leafref are used in grouping. In the leaf-list port-list, leafref type has been changed from leafref to string, so that the serialization can be performed. This is a workaround. Gilles raised a bug in JIRA, MDSAL-668 so that the initial issue can be solved.
The change 7.1 Renderer must be put on top of this change.
“Reduce the number of service notifications sent” simplifies the way we handle the MDONS use case.
“Add service listener to notify Kafka “ : this change allows sending a notification to a Kafka broker when a service is down. The listener shall be moved from the network model to the service handler, since it does not have any connection with the topology.
Javier mentioned that some code will be released soon for the notification service. He is also working on the functional tests (almost done for the T-API module)
Shweta proposed a change for deletion of B100G services : deleting the muxponder services implies considering cross-connect deletion.
Fix OLM timers : this change proposes to adjust timers and reduce the time associated with some of the operations.
OpenROADM version fix : after some discussion it was decided to keep it where it is in the relation chain. Moving it to the bottom would require a lot of work and does not bring value.
06/03/2021
General information
Next week will be the developer forum happening at the same time as OFC.
Si SR1 shall be approved today. We shall be ready by next Thursday.
Next week meeting will be canceled so that everybody can assist/participate to OFC and the LFN developer forum.
Code Review
Notification support for 7.1 device will allow to capture alarms for 7.1 devices. The change pushed on Master includes changes from Gilles.
Shweta pushed another change for switching-pool mapping update from notifications sent by equipment which solves a locking issue with the equipment that we had in previous version of the code.
Guillaume proposes to merge the changes as well as the one on Power control for 87.5 Ghz channels.
Guillaume merged portmapping functional test on master and cherry picked them on the Si branch
Bala will try to move Port mapping change to support Muxponder capability for B100G to the bottom of relation chain
Add missing if cap type for B100G could be merged.
We won’t have time to integrate Port Mapping change to support muxponders capability for B100G on Si SR1
Christophe pushed the functional test associated with the latest modifications brought in the PCE to handle OTUC4, ODUC4, and 100GE supported on ODUCN
The change “Reduce the number of service notifications sent” pushed by Thierry answers a request raised in the context of the MDONS use case to avoid the generation of unnecessary intermediate notifications.
Guillaume is currently working on automation to remove unneeded trailing spaces and tabs.
Gilles mentions we would have an interest in parallelizing tox jobs.
Guillaume agrees but mentions that we will have an issue as this can probably not be done right now with Jenkins Releng which relies on a project that is maintained by the LFN.
This means we would need to do this outside Jenkins Releng and need to manage the updates and maintain it ourselves.
Some solutions relying on python may be used but they are based on detox which is now obsolete.
Gilles can proceed with releasing tpce in Si SR1 as it is.
05/27/2021
05/27/21 is rescheduled at 2pm Paris Time to avoid overlap with OpenROADM meetings
General information
Silicon SR1 is in good shape.
Developer forum from the 7 to the 10Th of June is open to registration (free registration). Most sessions on ODL are planned on Tuesday with also some on ONAP. Unfortunately it is at the same time as OFC.
Code Review
Guillaume rebased OTN topology on top of the fix he provided.
The dependencies for Si SR1 have been bump. We should now use these ones (odl parent 8.1.2, Netconf 1.13.2, mdsal 7.07…).
Guillaume has pushed a change related to this (Bump dependencies to newer Silicon dev versions).
This change and the change “Fix new checkstyle issues found by odlparent 8.1.2“ shall be cherry-picked and merged on top of the relation chain for Si SR1 .
The change of Bala made on stable Si will also need to be rebased after this has been done.
“Add missing if-cap-type for B100G” : has been modified to consider the latest modifications made on other changes“Device renderer for Muxponder” : fixes a bug that was noticed during testing with lab equipment.“Distinguish flex vs fixed for PCE 400G” : also fixes a bug associated with central frequency handling.“Renderer support for higher rate” is adding support for OTUCn and ODUCn support in muxponder to create infrastructure connection type for OTN high rate services.