contexthelpers: Fix deadlock when nobody recvs on success channel #13408
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
brb
added
kind/bug
brb
force-pushed
the
pr/brb/fix-etcd-session-success
branch
from
4384690
to
9a697ed
Compare
It has been observed that the "sessionSuccess <- true" statement can
block forever. E.g. Cilium v1.7.9:
goroutine 742 [chan send, 547 minutes]:
github.com/cilium/cilium/pkg/kvstore.(*etcdClient).renewLockSession(0xc000f66000, 0x2790b40, 0xc000e247c0, 0x0, 0x0)
/go/src/github.com/cilium/cilium/pkg/kvstore/etcd.go:657 +0x3e2
github.com/cilium/cilium/pkg/kvstore.connectEtcdClient.func6(0x2790b40, 0xc000e247c0, 0x3aae820, 0x2)
/go/src/github.com/cilium/cilium/pkg/kvstore/etcd.go:819 +0x3e
github.com/cilium/cilium/pkg/controller.(*Controller).runController(0xc000f42500)
/go/src/github.com/cilium/cilium/pkg/controller/controller.go:205 +0xa2a
created by github.com/cilium/cilium/pkg/controller.(*Manager).updateController
/go/src/github.com/cilium/cilium/pkg/controller/manager.go:120 +0xb09
This can happen when the context cancellation timer fires after a
function which consumes the context has returned, but before
"sessionSuccess <- true" is executed. After the timeout, the receiving
goroutine is closed, making the sending block forever.
Fix this by making the sessionSuccess channel buffered. Note that after
sending we don't check whether the context has been cancelled, as we
expect that any subsequent functions which consume the context will
check for the cancellation.
Fixes: 0262854 ("etcd: Fix incorrect context usage in session renewal")
Signed-off-by: Martynas Pumputis <m@lambda.lt>
aanm
requested changes
Oct 5, 2020
brb
force-pushed
the
pr/brb/fix-etcd-session-success
branch
from
9a697ed
to
746674e
Compare
|
test-me-please |
aanm
approved these changes
Oct 6, 2020
gandro
approved these changes
Oct 6, 2020
tgraf
approved these changes
Oct 7, 2020
maintainer-s-little-helper
bot
added
the
ready-to-merge
maintainer-s-little-helper
bot
moved this from Needs backport from master
to Backport pending to v1.8
in 1.8.5
maintainer-s-little-helper
bot
moved this from Needs backport from master
to Backport pending to v1.6
in 1.6.13
maintainer-s-little-helper
bot
moved this from Needs backport from master
to Backport pending to v1.7
in 1.7.11
maintainer-s-little-helper
bot
moved this from Backport pending to v1.7
to Backport done to v1.7
in 1.7.11
maintainer-s-little-helper
bot
moved this from Backport pending to v1.6
to Backport done to v1.6
in 1.6.13
maintainer-s-little-helper
bot
moved this from Backport pending to v1.8
to Backport done to v1.8
in 1.8.5
maintainer-s-little-helper
bot
moved this from Backport pending to v1.8
to Backport done to v1.8
in 1.8.5
This was referenced Oct 28, 2020
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.
It has been observed that the
sessionSuccess <- truestatement can block forever. E.g. with Cilium v1.7.9:This can happen when the context cancellation timer fires after a function which consumes the context has returned, but before
sessionSuccess <- trueis executed. After the timeout, the receiving goroutine is closed, making the sending block forever.Fix this by making the
sessionSuccesschannel buffered. Note that after sending we don't check whether the context has been cancelled, as we expect that any subsequent functions which consume the context will check for the cancellation.Fixes: 0262854 ("etcd: Fix incorrect context usage in session renewal")