Fix deadlock issue around reaper's mutex during async-nsenter execution #36
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
Problem diagnosis has been well documented as part of issue [#266](nestybox/sysbox#266). Here I'm briefly explaining the fix that I've identified for this bug. As described in the issue's notes, the problem is caused by a deadlock scenario while dealing with the reaper's RWmutex. The fix is fairly simple: have 'async' nsenter logic remove its reaper's dependency. See that there's no real need for the reaper in the 'async' nsenter case as the SendRequest() callee will always sigkill() the nsenter process, so we can safely remove all this concurrency complexity in this case. Fix has been heavily tested over the last couple of weeks in the same scaling scenario where the problem was originally reported. No deadlock issues have been observed ever since. Signed-off-by: Rodny Molina <rmolina@nestybox.com>
ctalledo
reviewed
May 15, 2021
nsenter/event.go
Outdated
| @@ -548,6 +566,14 @@ func (e *NSenterEvent) ReceiveResponse() *domain.NSenterMessage { | |||
| return e.ResMsg | |||
| } | |||
|
|
|||
| func (e *NSenterEvent) ReapStartRequest() { | |||
There was a problem hiding this comment.
I don't see this new function called from anywhere; same for ReapEndRequest() below.
There was a problem hiding this comment.
You're right, i ended up not needing these new methods. Will remove them and merge.
ctalledo
reviewed
May 15, 2021
domain/nsenter.go
Outdated
| @@ -132,6 +132,8 @@ type NSenterEventIface interface { | |||
| SetResponseMsg(m *NSenterMessage) | |||
| GetResponseMsg() *NSenterMessage | |||
| GetProcessID() uint32 | |||
| ReapStartRequest() | |||
There was a problem hiding this comment.
I don't see any callers for these new interface methods.
ctalledo
approved these changes
May 15, 2021
Signed-off-by: Rodny Molina <rmolina@nestybox.com>
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.
Problem diagnosis has been well documented as part of issue #266. Here I'm briefly explaining the fix that I've identified for this bug.
As described in the issue's notes, the problem is caused by a deadlock scenario while dealing with the reaper's RWmutex. The fix is fairly simple: have 'async' nsenter logic remove its reaper's dependency. See that there's no real need for the reaper in the 'async' nsenter case as the SendRequest() callee will always sigkill() the nsenter process, so we can safely remove all this concurrency complexity in this case.
Fix has been heavily tested over the last couple of weeks in the same scaling scenario where the problem was originally reported. No deadlock issues has been observed ever since.
Signed-off-by: Rodny Molina rmolina@nestybox.com