Request each service's tasks separately #1489
Closed
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
There is currently a bug in swarmkit and the docker engine (moby/moby#37997) where large numbers of tasks can cause the message size to exceed the gRPC limits. A more appropriate fix involves changing the swarmkit gRPC api internally in some way or another to avoid this case occurring, but that fix requires more engineering effort and would take longer. As a short-term band-aid, when using the CLI, split task list requests by service, which will cause the message size to not grow beyond the gRPC limits. This fix leaks a bit of the abstraction presented by the engine API, but the resulting code is functionally equivalent. The alternative band-aid fix, contained solely to the engine, would involve attempting to deconstruct filters in the engine to make multiple requests, instead of merely converting. It is therefore cleaner to make this change here, where the request ultimately originates. This change can and will be reverted when the engine fix goes in. Signed-off-by: Drew Erny <drew.erny@docker.com>
Codecov Report
@@ Coverage Diff @@
## master #1489 +/- ##
==========================================
- Coverage 55.14% 54.23% -0.92%
==========================================
Files 289 289
Lines 19371 19361 -10
==========================================
- Hits 10683 10500 -183
- Misses 7997 8185 +188
+ Partials 691 676 -15 |
|
Just ran a test with this change and it definitely slows stuff down. With 17.06, on a system with about 70 services and about 20 tasks per service, takes about 16 seconds. With this PR, takes about 38 seconds. Thinking of alternatives... |
|
something's coming in the air tonight...hold on |
|
@dperny i elect to close this PR in favor of the work in moby to address the same issue: moby/moby#38103 |
|
Agreed, I don't think think this is the solution we're looking for 🤗 |
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.
Signed-off-by: Drew Erny drew.erny@docker.com
- What I did
When doing a
docker service ls, get each service's tasks in separate requests, instead of all in one request.- How I did it
There is currently a bug in swarmkit and the docker engine (moby/moby#37997) where large numbers of tasks can cause the message size to exceed the gRPC limits. A more appropriate fix involves changing the swarmkit gRPC api internally in some way or another to avoid this case occurring, but that fix requires more engineering effort and will take longer.
As a short-term band-aid, when using the CLI, split task list requests by service, which will cause the message size to not grow beyond the gRPC limits.
This fix leaks a bit of the abstraction presented by the engine API, but the resulting code is functionally equivalent. The alternative band-aid fix, contained solely to the engine, would involve attempting to deconstruct filters in the engine to make multiple requests, instead of merely converting. It is therefore cleaner to make this change here, where the request ultimately originates.
This change can and will be reverted when the engine fix goes in.
- How to verify it
As the code is functionally equivalent, the extant tests should cover it.
- Description for the changelog
Use a different request to get every service's tasks when doing a service ls to avoid a current engine bug.