A highly scalable redis-persistent queuing system for vert.x
- Clone this repository or unzip archive
- Install and start Redis
You need Java 11 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 |
processorTimeout | 240000 | The timeout [ms] to wait for the queue processor to answer the event bus message |
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 |
redisAuth | The authentication key (password) to connect to redis | |
maxPoolSize | 200 | The maximum size of the redis connection pool |
maxPoolWaitingSize | -1 | The maximum waiting requests for a connection from the pool |
maxPipelineWaitingSize | 2048 | The maximum allowed queued waiting handlers |
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. |
queueSpeedIntervalSec | 60 | The interval [s] to check queue speed |
memoryUsageLimitPercent | 100 | Percentage of the available system memory to be used by vertx-redisques. Only values between 0 and 100 are allowed. When the used memory ratio is higher than this limit, enqueues are rejected |
memoryUsageCheckIntervalSec | 60 | The interval [s] to check the current memory usage. memoryUsageCheckIntervalSec value must be greater 0, otherwise the default is used. |
redisReconnectAttempts | 0 | The amount of attempts to reconnect when redis connection is lost. Use 0 to not reconnect at all or -1 to reconnect indefinitely. |
redisReconnectDelaySec | 30 | The interval [s] to attempt to reconnect when redis connection is lost. |
redisPoolRecycleTimeoutMs | 180000 | The timeout [ms] when the connection pool is recycled. Use -1 when having reconnect feature enabled. |
micrometerMetricsEnabled | false | Enable / disable collection of metrics using micrometer |
micrometerMetricsIdentifier | default | Identifier to track values from multiple redisques instances |
httpRequestHandlerEnabled | false | Enable / disable the HTTP API |
httpRequestHandlerAuthenticationEnabled | false | Enable / disable authentication for the HTTP API |
httpRequestHandlerUsername | The username for the HTTP API authentication | |
httpRequestHandlerPassword | The password for the HTTP API authentication | |
enableQueueNameDecoding | true | Enable / disable the encoding of queue names when using 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 |
queueConfigurations | Configure retry intervals and enqueue delaying for queue patterns | |
dequeueStatisticReportIntervalSec | -1 | The interval [s] to publish the dequeue statistics into shared map. Use -1 to not publish at all. In a hazelcast-cluster environment need config Semaphore first, see: [Semaphore Config](#Semaphore Config) |
publish-metrics-address | The EventBus address to send collected redis metrics to | |
metric-storage-name | queue | The name of the storage used in the published metrics |
metric-refresh-period | 10 | The frequency [s] of collecting metrics from redis database |
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",
"payload": {
"filter": <str regex pattern to filter queues to count (optional)>
}
}
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",
"value": <long 1 when the queue was deleted, long 0 when no queue was found>
}
Request Data
{
"operation": "bulkDeleteQueues",
"payload": {
"queues": <JsonArray queues to delete>
}
}
Response Data
{
"status": "ok" / "error",
"value": <Long Amount of deleted queues>
}
Request Data
{
"operation": "getAllLocks",
"payload": {
"filter": <str regex pattern to filter locks (optional)>
}
}
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": "bulkPutLocks",
"payload": {
"locks": <JsonArray locks to add>,
"requestedBy": <str user who created the locks>
}
}
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"
}
Request Data
{
"operation": "bulkDeleteLocks",
"payload": {
"locks": <JsonArray locks to delete>
}
}
Response Data
{
"status": "ok" / "error",
"value": <Long Amount of deleted locks>
}
Request Data
{
"operation": "deleteAllLocks"
}
Response Data
{
"status": "ok" / "error",
"value": <Long Amount of deleted locks>
}
Request Data
{
"operation": "getQueueStatistics",
"payload": {
"filter": <str regex pattern to filter queues to retrieve statistics information (optional)>
}
}
Response Data
{
"status": "ok" / "error",
"queues": [ <JsonArray of resulting queue statistics>
{
"name": <str QUEUENAME>
"size": <Long Size of queue>
"failures": <Long Number of current failure count of queue>
"slowdownTime": <Long Current slowdown time in ms>
"backpressureTime": <Long Current backpressure time in ms>
}
]
}
Request Data
{
"operation": "getQueuesSpeed",
"payload": {
"filter": <str regex pattern to filter queues to retrieve the cumulated speed (optional)>
}
}
Response Data
{
"speed": <Long speed>
"unitSec": <Long seconds>
}
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. When authentication for the HTTP API is enabled, an Authorization request header (basic auth) has to be provided.
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
{
"address": "redisques",
"configuration-updated-address": "redisques-configuration-updated",
"redis-prefix": "redisques:",
"processor-address": "processor-address",
"refresh-period": 2,
"redisHost": "localhost",
"redisPort": 6379,
"redisReconnectAttempts": 0,
"redisReconnectDelaySec": 30,
"redisPoolRecycleTimeoutMs": 180000,
"redisAuth": null,
"checkInterval": 60,
"processorTimeout": 240000,
"processorDelayMax": 0,
"httpRequestHandlerEnabled": false,
"httpRequestHandlerPrefix": "/queuing",
"httpRequestHandlerPort": 7070,
"httpRequestHandlerUserHeader": "x-rp-usr",
"queueConfigurations": [{
"pattern": "queue.*",
"retryIntervals": [2, 7, 12, 17, 22, 27, 32, 37, 42, 47, 52],
"enqueueDelayFactorMillis": 0.0,
"enqueueMaxDelayMillis": 0
}],
"enableQueueNameDecoding": true,
"maxPoolSize": 200,
"maxPoolWaitingSize": -1,
"maxPipelineWaitingSize": 2048,
"queueSpeedIntervalSec": 60,
"memoryUsageLimitPercent": 100,
"memoryUsageCheckIntervalSec": 60
}
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
"processorTimeout": 1 // 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
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=true: 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=true: Lock the queue before enqueuing to prevent processing
When the locked=true 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
Available url parameters are:
- filter=: Filter the queues to list or count
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=true
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=true
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
The result will be a statusCode 200 OK when the queue could be successfully deleted or a 404 Not Found when the queue did not exist. Any error will result in a statusCode 500 Internal Server Error.
Available url parameters are:
- unlock=true: Unlock the queue after deleting all queue items
To delete a custom subset of existing queues use
POST /queuing/queues?bulkDelete=true
The payload must contain an array with the queues to delete.
Example:
{
"queues": [
"queue1",
"queue2"
]
}
The result will be a json object containing the number of deleted queues like the example below
{
"deleted": 2
}
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/
Available url parameters are:
- filter=: Filter the locks to return
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 add multiple locks use
POST /queuing/locks
The payload must contain an array with the locks to add.
Example:
{
"locks": [
"queue1",
"queue2"
]
}
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
To delete a custom subset of existing locks use
POST /queuing/locks?bulkDelete=true
The payload must contain an array with the locks to delete.
Example:
{
"locks": [
"queue1",
"queue2"
]
}
The result will be a json object containing the number of deleted locks like the example below
{
"deleted": 2
}
To delete all existing locks use
DELETE /queuing/locks
The result will be a json object containing the number of deleted locks like the example below
{
"deleted": 22
}
The statistics information contains the active queues and their queue items count plus the current number of failures and delay times in ms. To get the monitor information use
GET /queuing/statistics
Available url parameters are:
- limit: The maximum amount of queues to list
- emptyQueues=true: Also show empty queues
- filter=: Filter the queues to list or count
The result will be a json object with the statistics information like the example below.
{
"queues": [
{
"name": "queue_1",
"size": 3,
"failures": 0,
"slowdownTime": 0,
"backpressureTime": 0
}
]
}
The speed evaluation collects and calculates the cumulated speed of all queues matchting the given filter. To get the speed information use
GET /queuing/speed
Available url parameters are:
- filter=: Filter the queues for which the cumulated speed is evaluated
The result will be a json object with the speed of the last measurement period calculated over all queues matching the given filter regex. Additionally, the used measurement time in seconds is returned (eg. 60 seconds by default)
{
"speed": 42,
"unitSec": 60
}
If you are running in a Hazelcast cluster, the semaphore default permit must set to 1. https://github.com/vert-x3/vertx-hazelcast/blob/master/src/main/asciidoc/index.adoc#using-an-existing-hazelcast-cluster
SemaphoreConfig semaphoreConfig = new SemaphoreConfig().setInitialPermits(1).setJDKCompatible(false).setName("__vertx.*");
hazelcastConfig.getCPSubsystemConfig().addSemaphoreConfig(semaphoreConfig);
Besides the API, redisques provides some key metrics collected by micrometer.io.
The collected metrics include:
Metric name | Description |
---|---|
redisques_enqueue_success_total | Overall count of queue items to be enqueued |
redisques_enqueue_fail_total | Overall count of failing enqueues |
redisques_dequeue_total | Overall count of queue items to be dequeued from the queues |
redisques_active_queues | Overall count of active queues |
redisques_max_queue_size | Amount of queue items of the biggest queue |
When you include redisques in you project, you probably already have the configuration for publishing the metrics.
To export the metrics locally you have to add this dependency to the pom.xml
<dependency>
<groupId>io.micrometer</groupId>
<artifactId>micrometer-registry-prometheus</artifactId>
<version>${micrometer.version}</version>
</dependency>
Also add the micrometer configuration to RedisQuesRunner
class like this:
MicrometerMetricsOptions options = new MicrometerMetricsOptions()
.setPrometheusOptions(new VertxPrometheusOptions()
.setStartEmbeddedServer(true)
.setEmbeddedServerOptions(new HttpServerOptions().setPort(9101))
.setEnabled(true))
.setEnabled(true);
Vertx vertx = Vertx.vertx(new VertxOptions().setMetricsOptions(options));
Using the configuration above, the metrics can be accessed with
-
Starting from version 2.6.x redisques requires Java 11.
-
Redisques versions greater than 01.00.17 depend on Vert.x v3.2.0 and therefore require Java 8.