Fix missing routes caused by Node events coming out of order #1526
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for your PR. The following commands are available:
|
Codecov Report
@@ Coverage Diff @@
## master #1526 +/- ##
==========================================
+ Coverage 67.54% 67.59% +0.05%
==========================================
Files 169 169
Lines 13424 13471 +47
==========================================
+ Hits 9067 9106 +39
+ Misses 3416 3415 -1
- Partials 941 950 +9
Flags with carried forward coverage won't be shown. Click here to find out more.
|
|
/test-all |
Dyanngg
reviewed
Nov 10, 2020
| // event to be processed before proceeding, or the route installation and uninstallation operations may override or | ||
| // conflict with each other. | ||
| if len(nodesHaveSamePodCIDR) > 0 { | ||
| return fmt.Errorf("podCIDR %s for Node %s is duplicate with Node %s", node.Spec.PodCIDR, nodeName, nodesHaveSamePodCIDR[0].(*nodeRouteInfo).nodeName) |
There was a problem hiding this comment.
Should we mention that this error could be temporary? or something like addNoteRoute is deferred
There was a problem hiding this comment.
thanks for the suggestion, done
jianjuns
reviewed
Nov 10, 2020
| // event to be processed before proceeding, or the route installation and uninstallation operations may override or | ||
| // conflict with each other. | ||
| if len(nodesHaveSamePodCIDR) > 0 { | ||
| return fmt.Errorf("podCIDR %s for Node %s is duplicate with Node %s", node.Spec.PodCIDR, nodeName, nodesHaveSamePodCIDR[0].(*nodeRouteInfo).nodeName) |
antoninbas
reviewed
Nov 10, 2020
Merged
|
LGTM |
PodCIDRs can be released from deleted Nodes and allocated to new Nodes. For server side, it won't happen that a PodCIDR is allocated to more than one Node at any point. However, for client side, if a resync happens to occur when there are Node creation and deletion events, the informer will generate the events in a way that all creation events come before deletion ones even they actually happen in the opposite order on the server side. Therefore, a PodCIDR may appear in a new Node before the Node that previously owns it is removed. To ensure the stale routes, flows, and relevant cache of this podCIDR are removed appropriately, we wait for the Node deletion event to be processed before proceeding, or the route installation and uninstallation operations may override or conflict with each other.
tnqn
force-pushed
the
node-route-override
branch
from
915c4c8
to
f93cf30
Compare
antoninbas
approved these changes
Nov 11, 2020
|
/test-all |
|
/test-windows-conformance |
|
"TestHostPortPodConnectivity" in "Kind / E2e tests on a Kind cluster on Linux with Antrea NetworkPolicies enabled" failed because of "ErrImagePull", since the test has passed in other verifications, I will skip rerunning all kind tests. |
antoninbas
pushed a commit
to antoninbas/antrea
that referenced
this pull request
Nov 11, 2020
…io#1526) PodCIDRs can be released from deleted Nodes and allocated to new Nodes. For server side, it won't happen that a PodCIDR is allocated to more than one Node at any point. However, for client side, if a resync happens to occur when there are Node creation and deletion events, the informer will generate the events in a way that all creation events come before deletion ones even they actually happen in the opposite order on the server side. Therefore, a PodCIDR may appear in a new Node before the Node that previously owns it is removed. To ensure the stale routes, flows, and relevant cache of this podCIDR are removed appropriately, we wait for the Node deletion event to be processed before proceeding, or the route installation and uninstallation operations may override or conflict with each other.
antoninbas
pushed a commit
that referenced
this pull request
Nov 11, 2020
PodCIDRs can be released from deleted Nodes and allocated to new Nodes. For server side, it won't happen that a PodCIDR is allocated to more than one Node at any point. However, for client side, if a resync happens to occur when there are Node creation and deletion events, the informer will generate the events in a way that all creation events come before deletion ones even they actually happen in the opposite order on the server side. Therefore, a PodCIDR may appear in a new Node before the Node that previously owns it is removed. To ensure the stale routes, flows, and relevant cache of this podCIDR are removed appropriately, we wait for the Node deletion event to be processed before proceeding, or the route installation and uninstallation operations may override or conflict with each other.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PodCIDRs can be released from deleted Nodes and allocated to new Nodes.
For server side, it won't happen that a PodCIDR is allocated to more
than one Node at any point. However, for client side, if a resync
happens to occur when there are Node creation and deletion events, the
informer will generate the events in a way that all creation events come
before deletion ones even they actually happen in the opposite order on
the server side. Therefore, a PodCIDR may appear in a new Node before
the Node that previously owns it is removed.
To ensure the stale routes, flows, and relevant cache of this podCIDR
are removed appropriately, we wait for the Node deletion event to be
processed before proceeding, or the route installation and
uninstallation operations may override or conflict with each other.
Fixes #1527