[Bug] serverSocketBacklog is settable but the accept backlog is hard-coded to 1024 (regression of #3529)
Maintainers usually reply within 1 day
Assessment
- Difficulty
- 1/5
- Estimated time
- Under an hour
- Newbie friendliness
- 25/100
- Issue type
- Bug
- Clarity
- Clearly specified
- Activity status
- Stale
- Tech stack
- java
- Domain
- backend, networking
Research direction
The hard-coded ChannelOption.SO_BACKLOG value is in NettyRemotingServer#initServerBootstrap; compare it with the getter NettyServerConfig#getServerSocketBacklog(), which nothing in production reads. Done means the bootstrap option uses the configured backlog and NettyServerConfigTest still passes. A linked pull request (#11397) already targets this fix, so check it before starting.
Written by the indexing model from the issue text.
Description
Before Creating the Bug Report
- I found a bug, not just asking a question, which should be created in GitHub Discussions.
- I have searched the GitHub Issues and GitHub Discussions of this repository and believe that this is not a duplicate.
- I have confirmed that this bug belongs to the current repository, not other repositories of RocketMQ.
Runtime platform environment
OS: any (remoting module; broker / nameserver / proxy accept backlog)
RocketMQ version
branch: develop
Git commit id: a7edeecfb (2026-10-13; the files are unchanged by that commit)
JDK Version
Compiler: Temurin 1.8.0_504 (target 1.8, same as CI)
Describe the Bug
serverSocketBacklog is settable (field, getter/setter and the rocketmq.remoting.socket.backlog system property), but the
accept backlog is a hard-coded literal, so the option silently does nothing.
NettyServerConfig:
private int serverSocketBacklog = NettySystemConfig.socketBacklog;
...
public int getServerSocketBacklog() { return serverSocketBacklog; }
public void setServerSocketBacklog(int serverSocketBacklog) { this.serverSocketBacklog = serverSocketBacklog; }
NettyRemotingServer#initServerBootstrap:
.option(ChannelOption.SO_BACKLOG, 1024)
getServerSocketBacklog() has no production reader anywhere in the repository — the only caller is
NettyServerConfigTest, which asserts that the setter stores the value. This is a regression of #3528/#3529 ("The value of
ChannelOption.SO_BACKLOG can not be hard-coded", 2021), which made the bootstrap read this option; the RIP-31
BrokerContainer refactor reverted that single line to the literal while keeping the field, the getter and the system
property.
Impact: an operator raising the backlog for a broker/nameserver/proxy that must absorb connection storms gets no change at
all — the listen queue stays at 1024, so bursts can be refused (clients see connection resets/timeouts) despite the tuned
configuration, and the value reported back by the config dump looks as if it were in effect.
Steps to Reproduce
- Set
serverSocketBacklog=65535inbroker.conf(or-Drocketmq.remoting.socket.backlog=65535). - Start the broker and inspect the server bootstrap:
ChannelOption.SO_BACKLOGis1024, not65535. - A connection burst larger than 1024 pending accepts is refused by the OS, exactly as with the default.
Expected behavior
The bootstrap uses the configured backlog, as it did before the RIP-31 refactor.
Actual behavior
The literal 1024 is used; the option is inert.
Additional context
Fix: pass this.nettyServerConfig.getServerSocketBacklog() to the bootstrap option, as #3529 did. NettyRemotingServer is
the only place a listening socket is created inside this module, so the single line restores the option for broker,
nameserver and the proxy's remoting server.
- Dominant language
- Java
- Stars
- 22.7k
- Forks
- 12k
- Avg merge
- 3d 18h
- Merged PRs (30d)
- 17
Getting set up
- No Dockerfile or Docker Compose file
- Has a pull request template
- Read the contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from apache/rocketmq
-
[Bug] RemotingHelper.connect swaps the socket send and receive buffer configurationPossibly taken @amalsaidov claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
Maintainers usually reply within 1 day
-
CompressionType.of fails under the Turkish localePossibly taken @sharanggupta claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 83/100
Maintainers usually reply within 1 day
-
[rocketmq-client-python] TransactionMQProducer allocates an unused producer during initializationOpen
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
Maintainers usually reply within 1 day
-
[Bug] notifyMasterOnline aborts the long-polling wake-up with a NullPointerException when an arrival drains a bucket firstPossibly taken @iloveeyjafjalla claimed this 29 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
apache/rocketmq#11441 · 1 comment ·
Maintainers usually reply within 1 day
-
[Enhancement] FilterFactory duplicate-registration and unregister semantics lack testsPossibly taken @chennaji9 claimed this 34 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
apache/rocketmq#11340 · 2 comments ·
Maintainers usually reply within 1 day
Similar issues
-
bug emitter:client:csharp
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
microsoft/typespec#12572 · 1 comment · 1 reaction ·
Maintainers usually reply within 1 day
-
GCC 15.2: -Wcast-qual warning - cast discards const qualifier from pointer target type (H5trace.c)Open
Difficulty 2/5 1-3 hours Newbie friendliness 73/100
Maintainers usually reply within 1 day
-
needs-area-label
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 2 days
-
area/ai area/backend component/zeebe kind/task
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
spAnser/reacharound#38 ·