-
Notifications
You must be signed in to change notification settings - Fork 136
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
[WIP] Scale fixes - only LBG #814
Conversation
TODO: handle case when LBG are not supported by OVN. Signed-off-by: Dumitru Ceara <dceara@redhat.com>
Signed-off-by: Dumitru Ceara <dceara@redhat.com>
Signed-off-by: Dumitru Ceara <dceara@redhat.com>
Allow up to 1k pods per node. Signed-off-by: Dumitru Ceara <dceara@redhat.com>
Signed-off-by: Dumitru Ceara <dceara@redhat.com>
Signed-off-by: Dumitru Ceara <dceara@redhat.com>
…builds). By default log DB transactions for LSP and Port_Binding. In ovn-controller log iface-id-ver mismatches. Signed-off-by: Dumitru Ceara <dceara@redhat.com>
This includes ovsdb-idl changes from: dceara/ovs@0414012 Signed-off-by: Dumitru Ceara <dceara@redhat.com>
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: dceara The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
If a single routeInfo had multiple external routes that matched the same gateway IP, that routeInfo could be added to the returned list multiple times. It was only locked once, however. But deletePodGWRoutesForNamespace() iterates over the list returned by getRouteInfosForGateway() and calls Unlock() on each element in the list, which can create an unbalanced locking situation (multiple Unlocks for a single Lock) in the above situation where a routeInfo has multiple external routes matching the same gatewayIP. Signed-off-by: Dan Williams <dcbw@redhat.com>
At least give a clue what's wrong instead of just: "failed: exit status 1"
When deciding if a service needs updating, we do a reflect.DeepEqual on the existing load balancers that are stored in a cache, vs the load balancers that would be built by the operation. This comparison was incorrect, because it was comparing the UUID field, which will never be equal. The end result is potentially a lot more pressure on OVN for transactions that essentially do nothing. Signed-off-by: Tim Rozet <trozet@redhat.com>
There's no point in limiting the timeout to something lower than the CRI timeout; it just causes pods in large-scale clusters to hit the kubelet retry backoffs and take even longer to bring up. Continued improvements to OVN will reduce the need for longer timeouts in the near future. Signed-off-by: Dan Williams <dcbw@redhat.com>
This picks up OVN fix for setting lrp gw chassis: ovn-org/ovn@e6bcb88 Signed-off-by: Dumitru Ceara <dceara@redhat.com>
6855b4f
to
56b9506
Compare
/test images |
2 similar comments
/test images |
/test images |
/retest |
Cherry-picked from: ovn-org/ovn-kubernetes@f18feab Signed-off-by: Dan Williams <dcbw@redhat.com>
7586517
to
d8b7b0b
Compare
@dceara: PR needs rebase. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
@dceara: The following tests failed, say
Full PR test history. Your PR dashboard. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
Latest WIP scale testing PR: #839 |
cc: @tssurya