feature: Added uni tag usage #371
Conversation
|
E2E tests are passing:
|
|
Tag_type changed to string. Ignore error |
There was a problem hiding this comment.
It's shaping up well @Alopalao.
I encountered a traceback though (more details in the comments).
Also, I asked to address front-end, bump v3 on endpoints, and migration script too. Let me know what you think, also for v3, good that we have on our radar vlan range for 2023.2 too so that wouldn't require another bumping and we can consider it's expected for v3, OK?
- Updated readme - Updated changelog - Tag type integer compatibility returned. - No longer calling `notify_interface_tags()` from Interface
There was a problem hiding this comment.
@Alopalao, thanks for the updates. I closed many of the threads I opened. The only things remaining are:
- Enhance the DB script to also use commands and ideally provide aggregate|preview commands as you've done on your topology PR.
- Make the final decision regarding not bumping v3 and the compatibility that you're already in touch with @italovalcy (since this is a major decision that will require rework on existing PRs, let's try to settle this as soon as we can), thanks guys.
- Downgrade API urls version back to 2
|
Updated script and changelog. Also downgraded API versioning back to v2, commit c26527a89f2f0e0bd3186806033117f580991c46 |
There was a problem hiding this comment.
I was also simulating an upgrade with one intra-evpl and one inter-evpl, and the scripts executed idempotently as expected.
I noticed an unexpected side effect though which might be from PR #349, right after the OpenFlow handshake, it ended up activating the EVCs, without running sdntrace_cp (currently all activations after a restart are relying on sdntrace_cp, so I'm going to ask for you to reassess this part too, since it's like a pre-mature activation, notice that in the future though the state of EVC might be stored so this part might change again, @Ktmi is working on something related to this part, anyway, it's not immediately conflicting but something to be aware):
2023-09-27 14:08:16,770 - INFO [kytos.core.atcp_server] (MainThread) New connection from 127.0.0.1:33732
2023-09-27 14:08:16,771 - INFO [kytos.core.atcp_server] (MainThread) New connection from 127.0.0.1:33746
2023-09-27 14:08:16,772 - INFO [kytos.core.atcp_server] (MainThread) New connection from 127.0.0.1:33750
2023-09-27 14:08:17,277 - INFO [kytos.napps.kytos/of_core] (thread_pool_sb_1) Connection ('127.0.0.1', 33732), Switch 00:00:00:00:00:00:00:01: OPENFLOW HANDSHAKE COMPLETE
2023-09-27 14:08:17,280 - INFO [kytos.napps.kytos/of_core] (thread_pool_sb_1) Connection ('127.0.0.1', 33746), Switch 00:00:00:00:00:00:00:03: OPENFLOW HANDSHAKE COMPLETE
2023-09-27 14:08:17,284 - INFO [kytos.napps.kytos/of_core] (thread_pool_sb_0) Connection ('127.0.0.1', 33750), Switch 00:00:00:00:00:00:00:02: OPENFLOW HANDSHAKE COMPLETE
2023-09-27 14:08:17,351 - INFO [kytos.napps.kytos/mef_eline] (thread_pool_app_16) Activating EVC 136a09e1d2a240. Interfaces: {'00:00:00:00:00:00:00:01:1': {'status': 'UP', 'status_reason
': set()}, '00:00:00:00:00:00:00:01:2': {'status': 'UP', 'status_reason': set()}}.
2023-09-27 14:08:17,355 - INFO [kytos.napps.kytos/mef_eline] (thread_pool_app_16) Activating EVC 789c1eeeb1f74f. Interfaces: {'00:00:00:00:00:00:00:01:1': {'status': 'UP', 'status_reason
': set()}, '00:00:00:00:00:00:00:03:1': {'status': 'UP', 'status_reason': set()}}.
Also, as I mentioned on Slack to you, I noticed an unexpected redploy and also unexpected consistency check deleting flows. This didn't happen deterministically, so I'm not sure if it was a left over in my development environment, either way I'll recommend for you to explore and exercise the upgrade a bit more to confirm this part too:
2023-09-27 13:08:45,240 - INFO [kytos.napps.kytos/flow_manager] (thread_pool_sb_1) Consistency check: alien flow on switch 00:00:00:00:00:00:00:03
2023-09-27 13:08:45,241 - INFO [kytos.napps.kytos/flow_manager] (thread_pool_sb_1) Flow forwarded to switch 00:00:00:00:00:00:00:03 to be deleted. Flow: {'flows': [{'switch': '00:00:00:0
0:00:00:00:03', 'table_id': 0, 'match': {'in_port': 3, 'dl_vlan': 1}, 'priority': 20000, 'idle_timeout': 0, 'hard_timeout': 0, 'cookie': 12282418139759057986, 'id': '5c38f29eff18b71d1ac5
d779a1ca75f2', 'stats': {'byte_count': 0, 'duration_sec': 530, 'duration_nsec': 762000000, 'packet_count': 0}, 'cookie_mask': 0, 'instructions': [{'instruction_type': 'apply_actions', 'a
ctions': [{'action_type': 'pop_vlan'}, {'port': 1, 'action_type': 'output'}]}]}]}
2023-09-27 13:08:45,241 - INFO [kytos.napps.kytos/flow_manager] (thread_pool_sb_1) Consistency check: alien flow on switch 00:00:00:00:00:00:00:03
2023-09-27 13:08:45,242 - INFO [kytos.napps.kytos/flow_manager] (thread_pool_sb_1) Flow forwarded to switch 00:00:00:00:00:00:00:03 to be deleted. Flow: {'flows': [{'switch': '00:00:00:0
0:00:00:00:03', 'table_id': 0, 'match': {'in_port': 2, 'dl_vlan': 1}, 'priority': 20000, 'idle_timeout': 0, 'hard_timeout': 0, 'cookie': 12282418139759057986, 'id': 'c9e960a72d0dfa9fc18e
27165873eb1d', 'stats': {'byte_count': 0, 'duration_sec': 530, 'duration_nsec': 677000000, 'packet_count': 0}, 'cookie_mask': 0, 'instructions': [{'instruction_type': 'apply_actions', 'a
ctions': [{'action_type': 'pop_vlan'}, {'port': 1, 'action_type': 'output'}]}]}]}
2023-09-27 13:08:45,384 - INFO [uvicorn.access] (MainThread) 127.0.0.1:60194 - "POST /api/kytos/flow_manager/v2/flows/00%3A00%3A00%3A00%3A00%3A00%3A00%3A01 HTTP/1.1" 202
2023-09-27 13:08:45,390 - INFO [kytos.napps.kytos/flow_manager] (AnyIO worker thread) Send FlowMod from request dpid: 00:00:00:00:00:00:00:03, command: add, force: False, flows[0, 1]: [
{'match': {'in_port': 2, 'dl_vlan': 1}, 'cookie': 12282418139759057986, 'actions': [{'action_type': 'pop_vlan'}, {'action_type': 'output', 'port': 1}], 'owner': 'mef_eline', 'table_group
': 'evpl', 'table_id': 0, 'priority': 20000}]
2023-09-27 13:08:45,395 - INFO [uvicorn.access] (MainThread) 127.0.0.1:60208 - "POST /api/kytos/flow_manager/v2/flows/00%3A00%3A00%3A00%3A00%3A00%3A00%3A03 HTTP/1.1" 202
2023-09-27 13:08:45,401 - INFO [kytos.napps.kytos/mef_eline] (thread_pool_app_12) Failover path for EVC(73ea391b271c42, evpl) was deployed.
In terms of consistency check, I haven't encountered any unexpected diffs in the DB, so could be indeed a left over on OvS switches that I had, but let's see. Look forward to your feedback and exploratory test too.
|
Aldo, I reported the issue here #376, good thing it was not related to the new changes. I'll try to use a new topology and re-run the upgrade and finish the review here, and the other bug is addressed in another PR. Let's try to isolate things just so we can land this PR here soon if possible, let's see. |
There was a problem hiding this comment.
@Alopalao leaving this PR pre-approved.
Using another topology, I ended up hitting an early activation again, but that's not an issue from this PR, so it can be handled in a new PR with the linked issues from the previous comments:
2023-09-27 16:56:29,554 - INFO [kytos.napps.kytos/mef_eline] (thread_pool_app_1) Activating EVC ef632a9413b04c. Interfaces: {'00:00:00:00:00:00:00:01:1': {'status': 'UP', 'status_reason
': set()}, '00:00:00:00:00:00:00:01:2': {'status': 'UP', 'status_reason': set()}}.
2023-09-27 16:56:29,558 - INFO [kytos.napps.kytos/mef_eline] (thread_pool_app_1) Activating EVC d69490106b4344. Interfaces: {'00:00:00:00:00:00:00:01:1': {'status': 'UP', 'status_reason
': set()}, '00:00:00:00:00:00:00:03:1': {'status': 'UP', 'status_reason': set()}}.
- I tried out both scripts, they both worked idempotently, deleted evcs too,
interface_detailsavailable_tags ended up as expected.
If your exploratory tests are also passing and if you simulated other successful upgrades since e2e tests are passing I think we can land this PR (that way we can also land the other 3 related others too, facilitating for you to move forward with the epic), let us know.
Later on, please don't forget to also cover the GET endpoint on topology and augment e2e tests covering the new part (and also add new e2e tests with tag ranges on UNIs and confirming the correct allocation and when it shouldn't). cc'ing also @italovalcy and @Ktmi in case they want to review in the next days since they're involved with other mef_eline PRs too.
|
Let's ship it. @Alopalao check out the last comment. @italovalcy, btw, in the future when you have time your review is still appreciated, if there's any feedback to be addressed Aldo will discuss and take care in a next one, we'll prioritize merging this one. |
Closes #324
Closes #367
Summary
Delete deprecated
notify_link_available_tagsmethod fromutils.pyAdapted NNI tags to
vlan_poolepic contributions (kytos and topology PR)UNIs now use and make available tags when required (when created, updated, deleted)
Adapted NApp from
tag_typeinterger to stringLocal tests
Created, updated and deleted EVC, missing doing these actions in mass.
Uni test added
End-to-End Tests
E2E tests are passing, comment.