人生在世，看得透，又看得远者prevail everywhere. v.gd/YLHADG v.gd/cRqvfD v.gd/bKvfFH v.gd/sRm3tA: nsq

A realtime distributed messaging platform (forked from https://github.com/nsqio/nsq)

http://nsq.io/

Source: https://github.com/youzan/nsq
Issues: https://github.com/youzan/nsq/issues
Mailing List: nsq-users@googlegroups.com
IRC: #nsq on freenode
Docs: http://nsq.io
Twitter: @nsqio, @absolute8511

NSQ is a realtime distributed messaging platform designed to operate at scale, handling billions of messages per day.

It promotes distributed and decentralized topologies without single points of failure, enabling fault tolerance and high availability coupled with a reliable message delivery guarantee. See features & guarantees.

Operationally, NSQ is easy to configure and deploy (all parameters are specified on the command line and compiled binaries have no runtime dependencies). For maximum flexibility, it is agnostic to data format (messages can be JSON, MsgPack, Protocol Buffers, or anything else). Official Go and Python libraries are available out of the box (as well as many other client libraries) and, if you're interested in building your own, there's a protocol spec.

We publish binary releases for linux and darwin.

NOTE: master is our development branch and may not be stable at all times.

About this fork

This fork has been moved to https://github.com/youzan/nsq

This fork add many features missing in the original and has been used in Youzan.com. Features:

Replication
HA
Auto balance and migrate
Partition
Delivery in Order
Consume history data
Trace the message life-cycle
Performance improvement

Deploy Dependency

This fork need the etcd cluster, so to deploy we need set the configure for the nsqd and nsqlookupd. For details please refer to the sample configure files under contrib directory.

Other deploy can be the same as the Official.

有赞NSQ文档索引

用户指南

有赞NSQ用户指南

有赞NSQ部署运维指南

JavaSDK使用指南

内部实现文档

FAQ for this Fork

Can I use the Official client SDK?

Most client SDK can be used to communication with the new server without using the new features and for new features in this fork the new client SDK should be used.

Golang: https://github.com/youzan/go-nsq

Java: https://github.com/youzan/nsqJavaSDK

PHP: https://github.com/youzan/php-nsq-client

Auto topic create has been disabled.

In order to manager the topics by nsqlookupd, we removed the create api on the nsqd. So in order to create a new topic, you need create the topic in the nsqadmin or send the create api to the nsqlookupd as below:

curl -X POST "http://127.0.0.1:4161/topic/create?topic=xxx&partition_num=2&replicator=3&syncdisk=1000"

Explain:
partition_num=2 : topic partition number
replicator=3: topic replica number
syncdisk=1000: fsync disk every syncdisk messages
for ordered topic the query param "orderedmulti=true" is needed.

The new server introduced the partitions but in order to keep compatible with the old clients, each node can hold only one partition of the topic. So the partition_num*replicator should not be great than the nsqd nodes. For the ordered topic, we can create as much as we need partitions. (replicator <= number of nsqd node) Also, the first channel for this topic should be created after the topic is created, otherwise this topic is not allowed to be written.

The log can be adjust dynamically.

This is useful to debug on the fly.

nsqd: curl -X POST "http://127.0.0.1:4151/loglevel/set?loglevel=3"
nsqlookupd: curl -X POST "http://127.0.0.1:4161/loglevel/set?loglevel=3"

The log is much more detail when the log level increased.

Change the consume position to allow consume history or skip some messages.

curl -X POST -d "timestamp:xxxxxxxxx" "http://127.0.0.1:4151/channel/setoffset?topic=xxx&partition=xx&channel=xxx"

Post body explain:
timestamp:xxxx (the seconds since 1970-1-1)
virtual_queue:xxx (the queue offset position from the begin)
msgcount:xxx (the message count from the begin)

For multi nsqd nodes, each node for each partition should be done using the api above.