Comments (5)
I have done some more investigation, it seems this problem, happens for all errors. And has existed since at least 2.6.
RabbitMQ sends a channel.close message, but it doesn't actually close the damn channel.
I have created an example of this here: https://gist.github.com/9fe5507c2e7169bb93b2
from rabbitmq-server.
This is a bug in all of the client libraries I have used.
When a server sends a channel.close, the client is expected to acknowledge with a channel.closeOk. If the client does this the channel is properly released.
from rabbitmq-server.
My reading of AMQP Spec is that no CloseOk is required when a channel is closed with an exception.
I would suggest this a bug in the server code?
https://www.rabbitmq.com/resources/specs/amqp0-9-1.pdf
2.2.3 No Confirmations
4.5 Channel Closure
The server will consider a channel closed when any of these happen:
- Either peer closes the channel or the parent connection using a Close/Close-Ok handshake.
- Either peer raises an exception on the channel or the parent connection
from rabbitmq-server.
Since this thread is from 2012, keep a couple of things in mind:
- Things could have changed since then.
- The AMQP 0-9-1 spec has known issues, inconsistencies and statements open to interpretation. Team RabbitMQ has an entire errata document. It also has features that 10 year later can confidently be considered poorly thought out and in a couple of places RabbitMQ deviates to do what's more user-friendly.
- If a particular behaviour is demonstrated by most popular client libraries, including those our team maintains, it is extremely unlikely that it's going to change by now. If something is a truly annoying behaviour, it has been reported plenty of times in the ~ 10 years since the 0-9-1 spec was last revised.
RabbitMQ sends clients a channel.close-ok
in response to a channel.close
, so why would clients not be required send a response when RabbitMQ initiates a close?
Anyhow, a quick glance at rabbit_channel
/rabbit_reader
, Java client and Bunny suggest that
- A
channel.close-ok
is still required. - Java client and Bunny sent
channel.close-ok
as expected. - In Java client it has been the case since at least Jan 2011 (very likely way earlier, it's just as early as things go for
ChannelN
in the current codebase)
So the reporter probably hasn't used one of the clients maintained by our team. node-amqp
is known to be a terrible, buggy abandonware and we have been recommending avoiding it for years.
I cannot name a client that has caused more head scratching and false RabbitMQ bug accusations than node-amqp. Use amqp.node instead.
from rabbitmq-server.
Any further comments about how the spec should be interpreted belong to the mailing list.
from rabbitmq-server.
Related Issues (20)
- RabbitMQ 3.13.0 nodes with Consul peer discovery enabled fails to form a cluster HOT 1
- Odd characters in example file deps/rabbit/docs/rabbitmq.conf.example l.801 HOT 1
- Popup does not close in Managemet UI
- A `rabbitmq-queues grow` equivalent is missing for streams HOT 2
- Channel Metrics Cardinality - /metrics stops working HOT 3
- Unhandled function clause exception when retrieving queue for nonexistent virtual host HOT 2
- Add the ability to specify `$node` in log formatter format string
- HTTP API: GET /api/queues/{vhost}/{name} can return duplicate keys for quorum queues HOT 3
- 3.11: two Raft replicas are in the timeout state, one is a candidate HOT 5
- OpenID compliance check is not based on the final specification
- Possible race condition in classic queue deletion/declaration handling for multi-node clusters
- Quorum queue replica Erlang process names can conflict between queues in different virtual hosts HOT 2
- OpenID Connect RP-Initiated Logout should be optional HOT 1
- Optimise how rdq files are scanned at CQ shared message store recovery startup HOT 21
- Management UI creates classic queue instead of virtual host default queue type in some circumstances HOT 2
- When I create a stream, an error message "could not connect osiris to replica" appears in the log. HOT 1
- Make it possible to configure OpenId Connect endpoints rather than discover them dynamically via OpenId Connect Discovery endpoint
- 3.13.0 - 3.13.2: Dead-letter cycle detection can wrongly drop messages
- 3.13.0 - 3.13.2: Wrong warning messages that dead letter messages get dropped HOT 1
- Streams: consider allowing non-numerical values for publishingId HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from rabbitmq-server.