Cherry-pick to 5.2: Fix docker hanging when container killed #3634
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
No timeout was passed to the docker client. It seems in case of a killed container it can happen that the connection is hanging. To interrupt this connection, the timeout from the metricset is passed to the client. That means in case info for a container cannot be fetched, it will timeout. This change requires that the docker module is not run with a timeout of 3s seconds, which indirectly means a period of 3s. The reason is that already the http request waits ~2s for the response. So if 1s is set as timeout, all requests will timeout. Further changes: * Containers without names will be ignored, as these are containers for which the data could not be fetched. * Period was set to 1s by default instead of the period as document. This was changed. * Add documentation node about minimal period. Closes elastic#3610 The issue with this PR was introduce in 5.2.1 by fixing the memory leak. Before go routines just piled up, but now they caused filebeat to hang. This needs also backport to 5.2.2 (cherry picked from commit 99f17d6)
ruflin
force-pushed
the
backport_3612_5.2
branch
from
66003a7 to
4e8128d
Compare
leweafan
pushed a commit
to leweafan/beats
that referenced
this pull request
Apr 28, 2023
No timeout was passed to the docker client. It seems in case of a killed container it can happen that the connection is hanging. To interrupt this connection, the timeout from the metricset is passed to the client. That means in case info for a container cannot be fetched, it will timeout. This change requires that the docker module is not run with a timeout of 3s seconds, which indirectly means a period of 3s. The reason is that already the http request waits ~2s for the response. So if 1s is set as timeout, all requests will timeout. Further changes: * Containers without names will be ignored, as these are containers for which the data could not be fetched. * Period was set to 1s by default instead of the period as document. This was changed. * Add documentation node about minimal period. Closes elastic#3610 The issue with this PR was introduce in 5.2.1 by fixing the memory leak. Before go routines just piled up, but now they caused filebeat to hang. This needs also backport to 5.2.2 (cherry picked from commit 06ecd67)
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.
Cherry-pick of PR #3612 to 5.2 branch. Original message:
No timeout was passed to the docker client. It seems in case of a killed container it can happen that the connection is hanging. To interrupt this connection, the timeout from the metricset is passed to the client. That means in case info for a container cannot be fetched, it will timeout.
This change requires that the docker module is not run with a timeout of 3s seconds, which indirectly means a period of 3s. The reason is that already the http request waits ~2s for the response. So if 1s is set as timeout, all requests will timeout.
Further changes:
Closes #3610
The issue with this PR was introduce in 5.2.1 by fixing the memory leak. Before go routines just piled up, but now they caused filebeat to hang.
This needs also backport to 5.2.2