A highly scalable redis-persistent queuing system for vert.x.
- Clone this repository or unzip archive
- Install and start Redis
You need Java 8 and Maven.
cd vertx-redisques
mvn clean install
They are stored as redis lists, thus created only when needed and removed when empty. There is nothing left even if you just used thousands of queues with thousands of messages.
It is guaranteed that a queue is consumed by the same RedisQuesVerticle instance (a consumer). If no consumer is registered for a queue, one is assigned (this uses redis setnx to ensure only one is assigned). When idle for a given time, a consumer is removed. This prevents subscription leaks and makes recovering automatic when a consumer dies.
There is no single point of control/failure. Just create many instances of RedisQues, they will work together. If an instance dies, its queues will be assigned to other instances.
The following configuration values are available:
| Property | Default value | Description |
|---|---|---|
| address | redisques | The eventbus address the redisques module is listening to |
| configuration-updated-address | redisques-configuration-updated | The eventbus address the redisques module publishes the configuration updates to |
| redis-prefix | redisques: | Prefix for redis keys holding queues and consumers |
| processor-address | redisques-processor | Address of message processors |
| refresh-period | 10 | The frequency [s] of consumers refreshing their subscriptions to consume |
| processorDelayMax | 0 | The maximum delay [ms] to wait between queue items before notify the consumer |
| redisHost | localhost | The host where redis is running on |
| redisPort | 6379 | The port where redis is running on |
| redisEncoding | UTF-8 | The encoding to use in redis |
| checkInterval | 60 | The interval [s] to check timestamps of not-active / empty queues by executing check queue operation. checkInterval value must be greater 0, otherwise the default is used. |
| httpRequestHandlerEnabled | false | Enable / disable the HTTP API |
| httpRequestHandlerPrefix | /queuing | The url prefix for all HTTP API endpoints |
| httpRequestHandlerPort | 7070 | The port of the HTTP API |
| httpRequestHandlerUserHeader | x-rp-usr | The name of the header property where the user information is provided. Used for the HTTP API |
The configurations have to be passed as JsonObject to the module. For a simplified configuration the RedisquesConfigurationBuilder can be used.
Example:
RedisquesConfiguration config = RedisquesConfiguration.with()
.redisHost("anotherhost")
.redisPort(1234)
.build();
JsonObject json = config.asJsonObject();Properties not overridden will not be changed. Thus remaining default.
To use default values only, the RedisquesConfiguration constructor without parameters can be used:
JsonObject json = new RedisquesConfiguration().asJsonObject();Redisques API for Vert.x - Eventbus
Evenbus Settings:
address = redisque
For a simplified working with the Redisques module, see the RedisquesAPI class:
org.swisspush.redisques.util.RedisquesAPI
This class provides utility methods for a simple configuration of the queue operations. See Queue operations chapter below for details.
The following operations are available in the Redisques module.
Request Data
{
"operation": "getConfiguration"
}
Response Data
{
"status": "ok" / "error",
"value": <obj RESULT>
}
Request Data
{
"operation": "setConfiguration",
"payload": {
"<str propertyName>": <str propertyValue>,
"<str property2Name>": <str property2Value>,
"<str property3Name>": <str property3Value>
}
}
Response Data
{
"status": "ok" / "error",
"message": <string error message when status=error>
}
Request Data
{
"operation": "enqueue",
"payload": {
"queuename": <str QUEUENAME>
},
"message": {
"method": "POST",
"uri": <st REQUEST URI>,
"payload": null
}
}
Response Data
{
"status": "ok" / "error",
"message": "enqueued" / <str RESULT>
}
Request Data
{
"operation": "lockedEnqueue",
"payload": {
"queuename": <str QUEUENAME>,
"requestedBy": <str user who created the lock>
},
"message": {
"method": "POST",
"uri": <st REQUEST URI>,
"payload": null
}
}
Response Data
{
"status": "ok" / "error",
"message": "enqueued" / <str RESULT>
}
Request Data
{
"operation": "getQueues"
}
Response Data
{
"status": "ok" / "error",
"value": <objArr RESULT>
}
Request Data
{
"operation": "getQueuesCount"
}
Response Data
{
"status": "ok" / "error",
"value": <long RESULT>
}
Request Data
{
"operation": "getQueueItemsCount",
"payload": {
"queuename": <str QUEUENAME>
}
}
Response Data
{
"status": "ok" / "error",
"value": <long RESULT>
}
Request Data
{
"operation": "check"
}
Response Data
{}
Request Data
{
"operation": "reset"
}
Response Data
{}
Request Data
{
"operation": "stop"
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "getQueueItems",
"payload": {
"queuename": <str QUEUENAME>,
"limit": <str LIMIT>
}
}
Response Data
{
"status": "ok" / "error",
"value": <objArr RESULT>,
"info": <nbrArray with result array (value property) size and total queue item count (can be greater than limit)>
}
Request Data
{
"operation": "addQueueItem",
"payload": {
"queuename": <str QUEUENAME>,
"buffer": <str BUFFERDATA>
}
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "getQueueItem",
"payload": {
"queuename": <str QUEUENAME>,
"index": <int INDEX>
}
}
Response Data
{
"status": "ok" / "error",
"value": <obj RESULT>
}
Request Data
{
"operation": "replaceQueueItem",
"payload": {
"queuename": <str QUEUENAME>,
"buffer": <str BUFFERDATA>,
"index": <int INDEX>
}
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "deleteQueueItem",
"payload": {
"queuename": <str QUEUENAME>,
"index": <int INDEX>
}
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "deleteAllQueueItems",
"payload": {
"queuename": <str QUEUENAME>,
"unlock": true/false
}
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "getAllLocks"
}
Response Data
{
"status": "ok" / "error",
"value": <obj RESULT>
}
Request Data
{
"operation": "putLock",
"payload": {
"queuename": <str QUEUENAME>,
"requestedBy": <str user who created the lock>
}
}
Response Data
{
"status": "ok" / "error"
}
Request Data
{
"operation": "getLock",
"payload": {
"queuename": <str QUEUENAME>
}
}
Response Data
{
"status": "ok" / "error",
"value": <obj RESULT>
}
Request Data
{
"operation": "deleteLock",
"payload": {
"queuename": <str QUEUENAME>
}
}
Response Data
{
"status": "ok" / "error"
}
RedisQues provides a HTTP API to modify queues, queue items and get information about queue counts and queue item counts.
To enable the HTTP API, the httpRequestHandlerEnabled configuration property has to be set to TRUE.
For additional configuration options relating the HTTP API, see the Configuration section.
The following request examples will use a configured prefix of /queuing
To list the available endpoints use
GET /queuing
The result will be a json object with the available endpoints like the example below
{
"queuing": [
"locks/",
"queues/",
"monitor/",
"configuration/"
]
}The configuration information contains the currently active configuration values. To get the configuration use
GET /queuing/configuration
The result will be a json object with the configuration values like the example below
{
"redisHost": "localhost",
"checkInterval": 10,
"address": "redisques",
"configuration-updated-address": "redisques-configuration-updated",
"httpRequestHandlerEnabled": true,
"redis-prefix": "redisques:",
"processorTimeout": 240000,
"processorDelayMax": 0,
"refresh-period": 10,
"httpRequestHandlerPrefix": "/queuing",
"redisEncoding": "UTF-8",
"httpRequestHandlerPort": 7070,
"httpRequestHandlerUserHeader": "x-rp-usr",
"redisPort": 6379,
"processor-address": "redisques-processor"
}To set the configuration use
POST /queuing/configuration
having the payload in the request body. The current implementation supports the following configuration values only:
{
"processorDelayMax": 0 // number value in milliseconds
}
The following conditions will cause a 400 Bad Request response with a corresponding error message:
- Body is not a valid json object
- Body contains not supported configuration values
- Body does not contain the processorDelayMax property
The monitor information contains the active queues and their queue items count. To get the monitor information use
GET /queuing/monitor
Available url parameters are:
- limit: The maximum amount of queues to list
- emptyQueues: Also show empty queues
The result will be a json object with the monitor information like the example below
{
"queues": [
{
"name": "queue_1",
"size": 3
},
{
"name": "queue_2",
"size": 2
}
]
}To enqueue a new queue use
PUT /queuing/enqueue/myNewQueue
having the payload in the request body. When the request body is not a valid json object, a statusCode 400 with the error message 'Bad Request' will be returned.
Available url parameters are:
- locked: Lock the queue before enqueuing to prevent processing
When the locked url parameter is set, the configured httpRequestHandlerUserHeader property will be used to define the user which requested the lock. If no header is provided, "Unknown" will be used instead.
To list the active queues use
GET /queuing/queues
The result will be a json object with a list of active queues like the example below
{
"queues": [
"queue_1",
"queue_2",
"queue_3"
]
}Attention: The result will also contain empty queues when requested before the internal cleanup has passed. Use the monitor endpoint when non-empty queues should be listed only.
To get the count of active queues only, use
GET /queuing/queues?count
The result will be a json object with the count of active queues like the example below
{
"count": 3
}Attention: The count will also contain empty queues when requested before the internal cleanup has passed. Use the monitor endpoint when non-empty queues should be counted only.
To list the queue items of a single queue use
GET /queuing/queues/myQueue
Add the limit url parameter to define the maximum amount of queue items to retrieve
The result will be a json object with a list of queue items like the example below
{
"myQueue": [
"queueItem1",
"queueItem2",
"queueItem3",
"queueItem4"
]
}To get the count of queue items only, use
GET /queuing/queues/myQueue?count
The result will be a json object with the count of queue items like the example below
{
"count": 4
}To delete all queue items of a single queue use
DELETE /queuing/queues/myQueue
Available url parameters are:
- unlock: Unlock the queue after deleting all queue items
To get a single queue item use
GET /queuing/queues/myQueue/0
The result will be a json object representing the queue item at the given index (0 in the example above). When no queue item at the given index exists, a statusCode 404 with the error message 'Not Found' will be returned.
To replace a single queue item use
PUT /queuing/queues/myQueue/0
having the payload in the request body. The queue must be locked to perform this operation, otherwise a statusCode 409 with the error message 'Queue must be locked to perform this operation' will be returned. When no queue item at the given index exists, a statusCode 404 with the error message 'Not Found' will be returned.
To delete a single queue item use
DELETE /queuing/queues/myQueue/0
The queue must be locked to perform this operation, otherwise a statusCode 409 with the error message 'Queue must be locked to perform this operation' will be returned. When no queue item at the given index exists, a statusCode 404 with the error message 'Not Found' will be returned.
To add a queue item (to the end of the queue) use
POST /queuing/queues/myQueue/
having the payload in the request body. When the request body is not a valid json object, a statusCode 400 with the error message 'Bad Request' will be returned.
To list all existing locks use
GET /queuing/locks/
The result will be a json object with a list of all locks like the example below
{
"locks": [
"queue1",
"queue2"
]
}To add a lock use
PUT /queuing/locks/myQueue
having an empty json object {} in the body. The configured httpRequestHandlerUserHeader property will be used to define the user which requested the lock. If no header is provided, "Unknown" will be used instead.
To get a single lock use
GET /queuing/locks/queue1
The result will be a json object with the lock information like the example below
{
"requestedBy": "someuser",
"timestamp": 1478100330698
}To delete a single lock use
DELETE /queuing/locks/queue1
Redisques versions greater than 01.00.17 depend on Vert.x v3.2.0 and therefore require Java 8.
