fix(microservice): avoid rmq pingcheck leaving queues behind - #2736
Open
GiHoon1123 wants to merge 1 commit into
Open
fix(microservice): avoid rmq pingcheck leaving queues behind#2736GiHoon1123 wants to merge 1 commit into
GiHoon1123 wants to merge 1 commit into
Conversation
RabbitMQ's client-rmq transport asserts a queue on every connect()
unless noAssert is set, using RQM_DEFAULT_QUEUE / RQM_DEFAULT_QUEUE_OPTIONS
as the default when the caller doesn't specify a queue. Some versions
of @nestjs/microservices default that queue name to an empty string,
which tells RabbitMQ to generate a fresh server-side queue name on
every call - verified directly against a real broker: calling
channel.assertQueue('', {}) three times produces three distinct
amq.gen-* queues, each durable and not auto-deleted. Since a
health-check client reconnects on every pingCheck, this means every
health check run leaves one more orphaned durable queue behind
forever.
Set noAssert: true by default when the caller passes RMQ options
without specifying queue, noAssert, queueOptions, exchange or
routingKey themselves - a plain connectivity check has no reason to
declare a real queue at all. Any of those options being explicitly set
is treated as an opt-out, so this only changes the previously-undefined
default case.
Verified with a real RabbitMQ instance: after fixing, running the
existing 'RMQ should connect' e2e test and checking RabbitMQ's
management API afterward shows no queues left behind. Note: the
locally installed @nestjs/microservices is 11.0.11, whose default
queue name is still the literal string 'default' (idempotent to
reassert), not the empty-string default from later versions the
issue was filed against - so this fixes the mechanism directly
(confirmed against a real broker) without also reproducing
before/after against the exact regressed dependency version in one
environment.
Fixes: nestjs#2680
Author
|
Just checking in on this one when you have a chance. No rush. |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
PR Checklist
Please check if your PR fulfills the following requirements:
PR Type
What kind of change does this PR introduce?
What is the current behavior?
Issue Number: #2680
MicroserviceHealthIndicator.pingCheckwithTransport.RMQand noqueuespecified creates a new RabbitMQ connection and callsassertQueuewith whatever default queue name/options@nestjs/microservices' RMQ client falls back to. Depending on the version, that default queue name can be an empty string, which tells RabbitMQ to generate a fresh server-side queue name on every call. Verified this directly against a real broker: callingchannel.assertQueue('', {})three times produced three distinct queues (amq.gen-3T8sSiwFKQdPlXlClsU9gQ,amq.gen-uAM4xsfWRz42cY6qBmG7LQ,amq.gen-D5OdeGwtldapdHgI8p0I5A), eachdurable: true, auto_delete: false, exclusive: false- i.e. they persist in RabbitMQ forever once the health-check connection closes. Since a health check reconnects on everypingCheckcall, every health check tick leaves one more orphaned durable queue behind, eventually overloading RabbitMQ.What is the new behavior?
pingChecknow defaultsnoAssert: truefor RMQ options when the caller hasn't specifiedqueue,noAssert,queueOptions,exchange, orroutingKeythemselves. A plain connectivity check has no reason to declare a real queue at all, so this skips theassertQueuecall entirely for the common case. If any of those options are explicitly set, that's treated as an opt-out and the caller's configuration is used as-is, unchanged.Verification
What I did and didn't verify:
amqplibagainst a real broker, as described above.RMQ > should connecte2e test (uses default options, noqueuespecified) with this fix applied, then checked RabbitMQ's management API (/api/queues) afterward - empty, no queues left behind.queue/noAssert/queueOptions/exchange/routingKey, and leaving Kafka's existing defaults untouched) - covered by 6 new unit tests.@nestjs/microservicesversion the issue was filed against. The version installed in this repo is 11.0.11, whose default RMQ queue name is still the literal string'default'(asserting the same name repeatedly is idempotent, so it doesn't reproduce the queue accumulation on its own) rather than the empty-string default from the later version referenced in the issue. Verified the failure mechanism and the fix's effect independently instead of chaining them through the exact regressed dependency version in one run.Full test suite (8 files, 77 tests) and lint pass.
Does this PR introduce a breaking change?
The only behavior change is for RMQ health checks that don't already specify
queue/noAssert/queueOptions/exchange/routingKey- previously undefined/leaky behavior, not a documented contract.Other information