Use network buffer less aggressive in ping probe #634
Merged
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 @jumpojoy! This changes makes sense to me. Can you please sync your repo and send it again? There is a Dockerfile.dev in the PR causing conflicts. |
jumpojoy
force-pushed
the
fix_buffer_overflow
branch
from
6fc0728
to
0ef6a08
Compare
|
Thanks for review @manugarg, done. |
When have higher number of targets (seen on 600+), no matter in single probe or in different probes, we may cause network buffer overflow as we add packets to the buffer much faster than NIC can handle them. To avoid this we may increase net.core.wmem sysctl, or add 1 milisecond dalay between buffer writes (which is done by this patch). This allows to have stable results for even 2k targets in single probe. Also patch improves logging - Added debug log when probe is started. - Do not print duplicate p.name as it is always added by logger, add p.runCnt instead - Improve debug message for a case when packet does not match runId this is usually happening when we received packet after probe timeout.
jumpojoy
force-pushed
the
fix_buffer_overflow
branch
from
0ef6a08
to
2477b78
Compare
manugarg
reviewed
Nov 10, 2023
Co-authored-by: Manu Garg <manugarg@gmail.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Outdated suggestions cannot be applied.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.
When have higher number of targets (seen on 600+), no matter in single
probe or in different probes, we may cause network buffer overflow as
we add packets to the buffer much faster than NIC can handle them.
To avoid this we may increase net.core.wmem sysctl, or add 1 milisecond
dalay between buffer writes (which is done by this patch). This allows
to have stable results for even 2k targets in single probe.
Also patch improves logging
add p.runCnt instead
this is usually happening when we received packet after probe timeout.