Improve healthcheck responsiveness #579
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
lightswitch05
force-pushed
the
feature/improve-healthcheck-responsiveness
branch
4 times, most recently
from
270a791
to
cce4611
Compare
By default, dig will retry 2 times (for a total of 3 attempts) to get a response back. Each attempt defaults to 5 seconds. Before this change, a single docker healtcheck failure would really mean three failures and would take a total of 15 seconds before failing. By default, docker healthchecks will retry 3 times before considering a service unhealhy (with a 30 second interval). Combined with dig retries, this means it would take a total of 9 failed DNS responses before it considers the pihole to be unhealthy. Combining the retry between dig and docker, dig considers it a success if even 1/3 responses are recieved - and docker considers it a success if only 1/3 of those successes are successful. I'm not great at math - and order does make a difference - but I think that means as long as 1/9th of DNS queries are being answered - then docker thinks its healthy. Anyways, long story sort, dig doesn't need to have its own retry logic since docker already has a configuarable retry. I also disable recurse since the goal is to test this specific instance. Also removed duplicate import statement. Signed-off-by: Daniel <daniel@developerdan.com>
lightswitch05
force-pushed
the
feature/improve-healthcheck-responsiveness
branch
from
cce4611
to
cf6d74a
Compare
diginc
approved these changes
Feb 25, 2020
|
|
|
Thanks for the PR. I wasn't aware this was a problem. Nice fix. |
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Outdated suggestions cannot be applied.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.
Description
By default, dig will try 3 times to get a response back. Each attempt defaults to 5 seconds. Before this change, a single docker healtcheck failure would really mean three failures and would take a total of 15 seconds before failing. By default, docker healthchecks will retry 3 times before considering a service unhealhy (with a 30 second interval). Combined with dig retries, this means it would take a total of 9 failed DNS responses before it considers the pihole to be unhealthy. Combining the retry between dig and docker, dig considers it a success if even 1/3 responses are received - and docker considers it a success if only 1/3 of those attempts are successful. I'm not great at math - and order does make a difference - but I think that means as long as 1/9th of these queries are being answered - then docker thinks its healthy. Anyways, long story sort, dig doesn't need to have its own retry logic since docker already has a configurable retry. I also disable recurse since the goal is to test this specific instance.
Motivation and Context
I have not had issues with this. I'm just trying to get familiar with this project and this stood out to me as a small potential improvement.
How Has This Been Tested?
I tested this by running dig commands directly, as well as building the image locally. I was unable to figure out how to cause a running image to become unhealthy but remain running.
Types of changes
Checklist: