-
Notifications
You must be signed in to change notification settings - Fork 338
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Run ipv6 lane in control plane jobs #4106
Conversation
49de5db
to
695b06b
Compare
unrelated UT flake:
|
flake reported at #4031 |
we also need ipv6 or dualstack external-gateway jobs @tssurya |
e9aadbd
to
367e847
Compare
ok I rebased and now all the new tests that were added:
the multi-nic EIP ones are failing...ugh |
hitting:
on all EIP tests. lmc |
its related to:
|
367e847
to
9595263
Compare
fixed |
For EgressIPs the IPV6 infrastructure is not healthy. For example, I see healthchecks failing:
I will skip them for now but I am going to open cards to fix these |
LOL! EFW e2e's are failing with v6 for different reasons. In run https://github.com/ovn-org/ovn-kubernetes/actions/runs/7796371932/job/21261668937?pr=4106 if we take
So for some reason around 52 min mark we are not able to grep for the status in EFW. Should we be increasing the timeout period because in IC this just became asynchronous? I cannot find any useful logs in status-manager code that shows the status patch happening. Clearly something went wrong...
Will skip EFW and open a new issue |
9595263
to
01a5f9b
Compare
I don't get this:
https://github.com/ovn-org/ovn-kubernetes/actions/runs/7799278970/job/21270727244?pr=4106 |
91db673
to
53ccb16
Compare
saving flakes: https://github.com/ovn-org/ovn-kubernetes/actions/runs/7933330375/job/21683028471?pr=4106 |
53ccb16
to
4360406
Compare
man, LGW is flaking and timing out and exceeding the 3 hours..
what is this? all the unidling stuff is failing :/
|
CI seems slow...
This is where it starts:
|
@tssurya maybe we should identify these flakes that keep happening and skip them as well for now? |
yeah I agree
This set has failed twice now, I am skipping them and opening cards.. |
Run No4: https://github.com/ovn-org/ovn-kubernetes/actions/runs/7962558670/job/21809935569?pr=4106
OK its mostly the same things, EgressQoS on LGW: For sure something is broken here, these used to pass until last week |
Run No5: https://github.com/ovn-org/ovn-kubernetes/actions/runs/7962558670/job/21958818526?pr=4106 "2024-02-25T23:22:55.4292253Z Feb 25 23:22:55.428: INFO: Unable to fetch unidling-6501/execpod-noendpointssx8qp/agnhost-container logs: the server rejected our request for an unknown reason (get pods execpod-noendpointssx8qp)" |
latest run NO.6: https://github.com/ovn-org/ovn-kubernetes/actions/runs/7962558670/job/21973857074?pr=4106
this is already included in the next list of skips.. |
Should validate flow data of br-int is sent to an external gateway
is failing on IPV6+LGW lane
#4181
4360406
to
d24d41c
Compare
disabling ANP+ipv6 opened #4142 to track them |
Run No: 7 https://github.com/ovn-org/ovn-kubernetes/actions/runs/8053130165/job/21995745660?pr=4106#logs
|
d24d41c
to
795f895
Compare
This commit fixes things here and there in e2e's to make control planes run on v6 as well. Note that some tests are skipped since they cannot be run on github runners with v6 enabled. NOTE: One of the multicast test is moved from e2e.go to multicast.go where it belongs. Signed-off-by: Surya Seetharaman <suryaseetharaman.9@gmail.com>
795f895
to
8cf27a9
Compare
Last run was all green: https://github.com/ovn-org/ovn-kubernetes/actions/runs/8060967448/job/22018507506?pr=4106 |
all right second time green too! |
- What this PR does and why is it needed
This commit fixes things here and there
in e2e's to make control planes run on
v6 as well. Note that some tests are
skipped since they cannot be run on
github runners with v6 enabled.
NOTE: One of the multicast test is moved
from e2e.go to multicast.go where it belongs.
- Special notes for reviewersAdd support for IPV6 e2e's to run upstream
We must re-enable all the skipped tests: ci-ipv6
Supercedes #3512
What this PR does and why is it needed
I was trying to review Fixes Egress Firewall node selector for ipv6 #3503 (that got merged already so too late) but then was thinking why this wasn't caught in CI and @trozet pointed out that we are indeed not running v6 in control plane? LOL. I swear we used to.. Wondering at which point this changed, either ways don't want to git blame. Let's fix that.
Special notes for reviewers
NOTE: Some of these tests that check connectivity to internet cannot be run. See IPv6 on GitHub-hosted runners actions/runner-images#668 (comment) for details.
There were some past efforts to re-enable some of these skipped tests, but that never happened and they are
still failing v6 lane: Fix E2E IPv6 tests: adapt addresses and subnets #2505, Add IPv6 support for "Should validate ingress connectivity from an external gateway" e2e test #2524, Bump ovn version to 21.06.0-12.fc33 #2287; Enable IPv6 e2e tests for control-plane #2276; so going to skip them again.