diff --git a/README.md b/README.md index 4d82b3f..a7d365b 100644 --- a/README.md +++ b/README.md @@ -1,3 +1,7 @@ +# 📣 EOL Announcement + +IMPORTANT: This exporter only works with RabbitMQ 3. Please use the official exporter for RabbitMQ 4 or newer. See #383 for details. + # RabbitMQ Exporter [![Build Status](https://travis-ci.org/kbudde/rabbitmq_exporter.svg?branch=master)](https://travis-ci.org/kbudde/rabbitmq_exporter) [![Coverage Status](https://coveralls.io/repos/kbudde/rabbitmq_exporter/badge.svg?branch=master)](https://coveralls.io/r/kbudde/rabbitmq_exporter?branch=master) [![](https://images.microbadger.com/badges/image/kbudde/rabbitmq-exporter.svg)](http://microbadger.com/images/kbudde/rabbitmq-exporter "Get your own image badge on microbadger.com") Prometheus exporter for RabbitMQ metrics. @@ -93,182 +97,6 @@ following capabilities are currently supported in effective than pure-Erlang JSON encoding. So this greatly reduces monitoring overhead when we have a lot of objects in RabbitMQ. -**Note for users of rabbmitmq < 3.6** - -no_sort and bert are enabled by default. You must overwrite the default settings with: - - RABBIT_CAPABILITIES=nobert ./rabbitmq_exporter - -## Metrics - -All metrics (except golang/prometheus metrics) are prefixed with "rabbitmq_". - -### Global - -Always exported. - -metric | description --------| ------------ -|up | Was the last scrape of rabbitmq successful. -|module_up | Was the last scrape of rabbitmq module successful. labels: module -|module_scrape_duration_seconds | Duration of the last scrape of rabbitmq module. labels: module -|exporter_build_info | A metric with a constant '1' value labeled by version, revision, branch and build date on which the rabbitmq_exporter was built. - -### Overview - -Always exported. -Labels: cluster - -metric | description --------| ------------ -|channels | Number of channels| -|connections | Number of connections| -|consumers | Number of message consumers| -|queues | Number of queues in use| -|exchanges | Number of exchanges in use| -|queue_messages_global | Number ready and unacknowledged messages in cluster.| -|queue_messages_ready_global | Number of messages ready to be delivered to clients.| -|queue_messages_unacknowledged_global | Number of messages delivered to clients but not yet acknowledged.| -|version_info| A metric with a constant '1' value labeled by rabbitmq version, erlang version, node, cluster.| - -### Queues - -Labels: cluster, vhost, queue, durable, policy, self - -#### Queues - Gauge - -metric | description --------| ------------ -|queue_messages_ready|Number of messages ready to be delivered to clients.| -|queue_messages_unacknowledged|Number of messages delivered to clients but not yet acknowledged.| -|queue_messages|Sum of ready and unacknowledged messages (queue depth).| -|queue_messages_ack_total|Number of messages delivered in acknowledgement mode in response to basic.get.| -|queue_messages_ready_ram|Number of messages from messages_ready which are resident in ram.| -|queue_messages_unacknowledged_ram|Number of messages from messages_unacknowledged which are resident in ram.| -|queue_messages_ram|Total number of messages which are resident in ram.| -|queue_messages_persistent|Total number of persistent messages in the queue (will always be 0 for transient queues).| -|queue_message_bytes|Sum of the size of all message bodies in the queue. This does not include the message properties (including headers) or any overhead.| -|queue_message_bytes_ready|Like message_bytes but counting only those messages ready to be delivered to clients.| -|queue_message_bytes_unacknowledged|Like message_bytes but counting only those messages delivered to clients but not yet acknowledged.| -|queue_message_bytes_ram|Like message_bytes but counting only those messages which are in RAM.| -|queue_message_bytes_persistent|Like message_bytes but counting only those messages which are persistent.| -|queue_consumers|Number of consumers.| -|queue_consumer_utilisation|Fraction of the time (between 0.0 and 1.0) that the queue is able to immediately deliver messages to consumers. This can be less than 1.0 if consumers are limited by network congestion or prefetch count.| -|queue_memory|Bytes of memory consumed by the Erlang process associated with the queue, including stack, heap and internal structures.| -|queue_head_message_timestamp|The timestamp property of the first message in the queue, if present. Timestamps of messages only appear when they are in the paged-in state.| -|queue_max_length_bytes|Total body size for ready messages a queue can contain before it starts to drop them from its head.| -|queue_max_length|How many (ready) messages a queue can contain before it starts to drop them from its head.| -|queue_idle_since_seconds|starttime where the queue switched to idle state; seconds since epoch (1970); only set if queue state is idle| -|queue_reductions_total|Number of reductions which take place on this process.| -|queue_state|A metric with a value of constant '1' if the queue is in a certain state. Labels: vhost, queue, *state* (running, idle, flow,..)| -|queue_slave_nodes_len|Number of slave nodes attached to the queue| -|queue_synchronised_slave_nodes_len|Number of slave nodes in sync to the queue| - -#### Queues - Counter - -metric | description --------| ------------ -|queue_disk_reads_total|Total number of times messages have been read from disk by this queue since it started.| -|queue_disk_writes_total|Total number of times messages have been written to disk by this queue since it started.| -|queue_messages_published_total|Count of messages published.| -|queue_messages_confirmed_total|Count of messages confirmed. | -|queue_messages_delivered_total|Count of messages delivered in acknowledgement mode to consumers.| -|queue_messages_delivered_noack_total|Count of messages delivered in no-acknowledgement mode to consumers. | -|queue_messages_get_total|Count of messages delivered in acknowledgement mode in response to basic.get.| -|queue_messages_get_noack_total|Count of messages delivered in no-acknowledgement mode in response to basic.get.| -|queue_messages_redelivered_total|Count of subset of messages in deliver_get which had the redelivered flag set.| -|queue_messages_returned_total|Count of messages returned to publisher as unroutable.| - -### Exchanges - Counter - -Labels: cluster, vhost, exchange - -metric | description --------| ------------ -|exchange_messages_published_in_total|Count of messages published in to an exchange, i.e. not taking account of routing.| -|exchange_messages_published_out_total|Count of messages published out of an exchange, i.e. taking account of routing.| - -### Node - Counter - -Labels: cluster, node, self - -metric | description --------| ------------ -|uptime|Uptime in milliseconds| -|running|number of running nodes| -|node_mem_used|Memory used in bytes| -|node_mem_limit|Point at which the memory alarm will go off| -|node_mem_alarm|Whether the memory alarm has gone off| -|node_disk_free|Disk free space in bytes.| -|node_disk_free_alarm|Whether the disk alarm has gone off.| -|node_disk_free_limit|Point at which the disk alarm will go off.| -|fd_used|Used File descriptors| -|fd_available|File descriptors available| -|sockets_used|File descriptors used as sockets.| -|sockets_available|File descriptors available for use as sockets| -|partitions | Current Number of network partitions. 0 is ok. If the cluster is splitted the value is at least 2| - -### Connections - Gauge - -_disabled by default_. Depending on the environment and change rate it can create a high number of dead metrics. Otherwise it could be usefull and can be enabled. - -Labels: cluster, vhost, node, peer_host, user, self - -Please note: The data is aggregated by label values as it is possible that there are multiple connections for a certain combination of labels. - -metric | description --------| ------------ -|connection_channels|number of channels in use| -|connection_received_bytes|received bytes| -|connection_received_packets|received packets| -|connection_send_bytes|send bytes| -|connection_send_packets|send packets| -|connection_send_pending|Send queue size| - -Labels: vhost, node, peer_host, user, *state* (running, flow,..), self - -metric | description --------| ------------ -|connection_status|Number of connections in a certain state aggregated per label combination. Metric will disappear if there are no connections in a state. | - -### Shovel - -_disabled by default_ -Labels: cluster, vhost, shovel, type, self, state - -metric | description --------| ------------ -|shovel_state|A metric with a value of constant '1' for each shovel in a certain state| - -### Memory - -_disabled by default_ -Labels: cluster, node, self - -metric | description --------| ------------ -|memory_allocated_unused_bytes|Memory preallocated by the runtime (VM allocators) but not yet used| -|memory_atom_bytes|Memory used by atoms. Should be fairly constant| -|memory_binary_bytes|Memory used by shared binary data in the runtime. Most of this memory is message bodies and metadata.)| -|memory_code_bytes|Memory used by code (bytecode, module metadata). This section is usually fairly constant and relatively small (unless the node is entirely blank and stores no data).| -|memory_connection_channels_bytes|Memory used by client connections - channels| -|memory_connection_other_bytes|Memory used by client connection - other| -|memory_connection_readers|Memory used by processes responsible for connection parser and most of connection state. Most of their memory attributes to TCP buffers| -|memory_connection_writers_bytes|Memory used by processes responsible for serialization of outgoing protocol frames and writing to client connections socktes| -|memory_mgmt_db_bytes|Management DB ETS tables + processes| -|memory_mnesia_bytes|Internal database (Mnesia) tables keep an in-memory copy of all its data (even on disc nodes)| -|memory_msg_index_bytes|Message index ETS + processes| -|memory_other_ets_bytes|Other in-memory tables besides those belonging to the stats database and internal database tables| -|memory_other_proc_bytes|Memory used by all other processes that RabbitMQ cannot categorise| -|memory_other_system_bytes|Memory used by all other system that RabbitMQ cannot categorise| -|memory_plugins_bytes|Memory used by plugins (apart from the Erlang client which is counted under Connections, and the management database which is counted separately).| -|memory_queue_procs_bytes|Memory used by class queue masters, queue indices, queue state| -|memory_queue_slave_procs_bytes|Memory used by class queue mirrors, queue indices, queue state| -|memory_reserved_unallocated_bytes|Memory preallocated/reserved by the kernel but not the runtime| -|memory_total_allocated_bytes|Node-local total memory - allocated| -|memory_total_rss_bytes|Node-local total memory - rss| -|memory_total_erlang_bytes|Node-local total memory - erlang| - ## Comparison to official exporter [official exporter](https://www.rabbitmq.com/prometheus.html): @@ -291,10 +119,32 @@ probalby best solution is to use both exporters: ### msg: Error while retrieving data from rabbitHost statusCode: 500 This exporter expects capabilities from rabbitmq 3.6.8 or newer by default. -If you are running older than 3.6.8 you have to set RABBIT_CAPABILITIES=compat. +If you are running older than 3.6.8 you must disable bert and no_sort with the setting RABBIT_CAPABILITIES=compat. +If you are running 3.13.0 or newer you must disable no_sort with the setting RABBIT_CAPABILITIES=no_sort. ### missing data in graphs -If there is a load balancer between the exporter and the RabbitMQApi, the setting `RABBIT_CONNECTION=loadbalancer` must be activated. +If there is a load balancer between the exporter and the RabbitMQApi, the setting `RABBIT_CONNECTION=loadbalancer` must be activated. See https://github.com/kbudde/rabbitmq_exporter/issues/131 for details. +## build and test + +This project uses goreleaser to build and release the project. You can build the project with the following command: + + goreleaser build --snapshot + +`go build` will also work, but it will not include the version information, uses cgo, etc. + +To run the tests, use the following command: + + go test -v ./... + +If you have docker installed, you can run the tests with the following command: + + go test -v ./... --tags integration + +This will start a rabbitmq container and run the tests against it. + +## Metrics + +The metrics are documented in the [metrics.md](metrics.md) file. diff --git a/integration_test.go b/integration_test.go index bf6d3db..e138fb6 100644 --- a/integration_test.go +++ b/integration_test.go @@ -31,7 +31,7 @@ func TestQueueCount(t *testing.T) { exporterURL = fmt.Sprintf("http://localhost:%s/metrics", defaultConfig.PublishPort) rabbitManagementURL = env.ManagementURL() os.Setenv("RABBIT_URL", rabbitManagementURL) - os.Setenv("RABBIT_CAPABILITIES", "bert,no_sort") + os.Setenv("RABBIT_CAPABILITIES", "bert") // no_sort not supported in rabbitmq 3.13+ defer os.Unsetenv("RABBIT_URL") defer os.Unsetenv("RABBIT_CAPABILITIES") diff --git a/metrics.md b/metrics.md new file mode 100644 index 0000000..e81b130 --- /dev/null +++ b/metrics.md @@ -0,0 +1,169 @@ +# Metrics + +All metrics (except golang/prometheus metrics) are prefixed with "rabbitmq_". + +## Global + +Always exported. + +metric | description +-------| ------------ +|up | Was the last scrape of rabbitmq successful. +|module_up | Was the last scrape of rabbitmq module successful. labels: module +|module_scrape_duration_seconds | Duration of the last scrape of rabbitmq module. labels: module +|exporter_build_info | A metric with a constant '1' value labeled by version, revision, branch and build date on which the rabbitmq_exporter was built. + +## Overview + +Always exported. +Labels: cluster + +metric | description +-------| ------------ +|channels | Number of channels| +|connections | Number of connections| +|consumers | Number of message consumers| +|queues | Number of queues in use| +|exchanges | Number of exchanges in use| +|queue_messages_global | Number ready and unacknowledged messages in cluster.| +|queue_messages_ready_global | Number of messages ready to be delivered to clients.| +|queue_messages_unacknowledged_global | Number of messages delivered to clients but not yet acknowledged.| +|version_info| A metric with a constant '1' value labeled by rabbitmq version, erlang version, node, cluster.| + +## Queues + +Labels: cluster, vhost, queue, durable, policy, self + +### Queues - Gauge + +metric | description +-------| ------------ +|queue_messages_ready|Number of messages ready to be delivered to clients.| +|queue_messages_unacknowledged|Number of messages delivered to clients but not yet acknowledged.| +|queue_messages|Sum of ready and unacknowledged messages (queue depth).| +|queue_messages_ack_total|Number of messages delivered in acknowledgement mode in response to basic.get.| +|queue_messages_ready_ram|Number of messages from messages_ready which are resident in ram.| +|queue_messages_unacknowledged_ram|Number of messages from messages_unacknowledged which are resident in ram.| +|queue_messages_ram|Total number of messages which are resident in ram.| +|queue_messages_persistent|Total number of persistent messages in the queue (will always be 0 for transient queues).| +|queue_message_bytes|Sum of the size of all message bodies in the queue. This does not include the message properties (including headers) or any overhead.| +|queue_message_bytes_ready|Like message_bytes but counting only those messages ready to be delivered to clients.| +|queue_message_bytes_unacknowledged|Like message_bytes but counting only those messages delivered to clients but not yet acknowledged.| +|queue_message_bytes_ram|Like message_bytes but counting only those messages which are in RAM.| +|queue_message_bytes_persistent|Like message_bytes but counting only those messages which are persistent.| +|queue_consumers|Number of consumers.| +|queue_consumer_utilisation|Fraction of the time (between 0.0 and 1.0) that the queue is able to immediately deliver messages to consumers. This can be less than 1.0 if consumers are limited by network congestion or prefetch count.| +|queue_memory|Bytes of memory consumed by the Erlang process associated with the queue, including stack, heap and internal structures.| +|queue_head_message_timestamp|The timestamp property of the first message in the queue, if present. Timestamps of messages only appear when they are in the paged-in state.| +|queue_max_length_bytes|Total body size for ready messages a queue can contain before it starts to drop them from its head.| +|queue_max_length|How many (ready) messages a queue can contain before it starts to drop them from its head.| +|queue_idle_since_seconds|starttime where the queue switched to idle state; seconds since epoch (1970); only set if queue state is idle| +|queue_reductions_total|Number of reductions which take place on this process.| +|queue_state|A metric with a value of constant '1' if the queue is in a certain state. Labels: vhost, queue, *state* (running, idle, flow,..)| +|queue_slave_nodes_len|Number of slave nodes attached to the queue| +|queue_synchronised_slave_nodes_len|Number of slave nodes in sync to the queue| + +### Queues - Counter + +metric | description +-------| ------------ +|queue_disk_reads_total|Total number of times messages have been read from disk by this queue since it started.| +|queue_disk_writes_total|Total number of times messages have been written to disk by this queue since it started.| +|queue_messages_published_total|Count of messages published.| +|queue_messages_confirmed_total|Count of messages confirmed. | +|queue_messages_delivered_total|Count of messages delivered in acknowledgement mode to consumers.| +|queue_messages_delivered_noack_total|Count of messages delivered in no-acknowledgement mode to consumers. | +|queue_messages_get_total|Count of messages delivered in acknowledgement mode in response to basic.get.| +|queue_messages_get_noack_total|Count of messages delivered in no-acknowledgement mode in response to basic.get.| +|queue_messages_redelivered_total|Count of subset of messages in deliver_get which had the redelivered flag set.| +|queue_messages_returned_total|Count of messages returned to publisher as unroutable.| + +## Exchanges - Counter + +Labels: cluster, vhost, exchange + +metric | description +-------| ------------ +|exchange_messages_published_in_total|Count of messages published in to an exchange, i.e. not taking account of routing.| +|exchange_messages_published_out_total|Count of messages published out of an exchange, i.e. taking account of routing.| + +## Node - Counter + +Labels: cluster, node, self + +metric | description +-------| ------------ +|uptime|Uptime in milliseconds| +|running|number of running nodes| +|node_mem_used|Memory used in bytes| +|node_mem_limit|Point at which the memory alarm will go off| +|node_mem_alarm|Whether the memory alarm has gone off| +|node_disk_free|Disk free space in bytes.| +|node_disk_free_alarm|Whether the disk alarm has gone off.| +|node_disk_free_limit|Point at which the disk alarm will go off.| +|fd_used|Used File descriptors| +|fd_available|File descriptors available| +|sockets_used|File descriptors used as sockets.| +|sockets_available|File descriptors available for use as sockets| +|partitions | Current Number of network partitions. 0 is ok. If the cluster is splitted the value is at least 2| + +## Connections - Gauge + +_disabled by default_. Depending on the environment and change rate it can create a high number of dead metrics. Otherwise it could be usefull and can be enabled. + +Labels: cluster, vhost, node, peer_host, user, self + +Please note: The data is aggregated by label values as it is possible that there are multiple connections for a certain combination of labels. + +metric | description +-------| ------------ +|connection_channels|number of channels in use| +|connection_received_bytes|received bytes| +|connection_received_packets|received packets| +|connection_send_bytes|send bytes| +|connection_send_packets|send packets| +|connection_send_pending|Send queue size| + +Labels: vhost, node, peer_host, user, *state* (running, flow,..), self + +metric | description +-------| ------------ +|connection_status|Number of connections in a certain state aggregated per label combination. Metric will disappear if there are no connections in a state. | + +## Shovel + +_disabled by default_ +Labels: cluster, vhost, shovel, type, self, state + +metric | description +-------| ------------ +|shovel_state|A metric with a value of constant '1' for each shovel in a certain state| + +## Memory + +_disabled by default_ +Labels: cluster, node, self + +metric | description +-------| ------------ +|memory_allocated_unused_bytes|Memory preallocated by the runtime (VM allocators) but not yet used| +|memory_atom_bytes|Memory used by atoms. Should be fairly constant| +|memory_binary_bytes|Memory used by shared binary data in the runtime. Most of this memory is message bodies and metadata.)| +|memory_code_bytes|Memory used by code (bytecode, module metadata). This section is usually fairly constant and relatively small (unless the node is entirely blank and stores no data).| +|memory_connection_channels_bytes|Memory used by client connections - channels| +|memory_connection_other_bytes|Memory used by client connection - other| +|memory_connection_readers|Memory used by processes responsible for connection parser and most of connection state. Most of their memory attributes to TCP buffers| +|memory_connection_writers_bytes|Memory used by processes responsible for serialization of outgoing protocol frames and writing to client connections socktes| +|memory_mgmt_db_bytes|Management DB ETS tables + processes| +|memory_mnesia_bytes|Internal database (Mnesia) tables keep an in-memory copy of all its data (even on disc nodes)| +|memory_msg_index_bytes|Message index ETS + processes| +|memory_other_ets_bytes|Other in-memory tables besides those belonging to the stats database and internal database tables| +|memory_other_proc_bytes|Memory used by all other processes that RabbitMQ cannot categorise| +|memory_other_system_bytes|Memory used by all other system that RabbitMQ cannot categorise| +|memory_plugins_bytes|Memory used by plugins (apart from the Erlang client which is counted under Connections, and the management database which is counted separately).| +|memory_queue_procs_bytes|Memory used by class queue masters, queue indices, queue state| +|memory_queue_slave_procs_bytes|Memory used by class queue mirrors, queue indices, queue state| +|memory_reserved_unallocated_bytes|Memory preallocated/reserved by the kernel but not the runtime| +|memory_total_allocated_bytes|Node-local total memory - allocated| +|memory_total_rss_bytes|Node-local total memory - rss| +|memory_total_erlang_bytes|Node-local total memory - erlang| diff --git a/testenv/Dockerfile b/testenv/Dockerfile index a3a72b1..89c00e1 100644 --- a/testenv/Dockerfile +++ b/testenv/Dockerfile @@ -1,2 +1,2 @@ -FROM rabbitmq:3.12-management-alpine -RUN rabbitmq-plugins enable --offline rabbitmq_shovel rabbitmq_shovel_management \ No newline at end of file +FROM rabbitmq:3.13-management-alpine +RUN rabbitmq-plugins enable --offline rabbitmq_shovel rabbitmq_shovel_management diff --git a/testenv/testenv.go b/testenv/testenv.go index 84126d1..08e6cab 100644 --- a/testenv/testenv.go +++ b/testenv/testenv.go @@ -18,7 +18,7 @@ import ( dockertest "github.com/ory/dockertest/v3" ) -//list of docker tags with rabbitmq versions +// list of docker tags with rabbitmq versions const ( RabbitMQ3Latest = "3-management-alpine" ) @@ -26,7 +26,7 @@ const ( // MaxWait is time before the docker setup will fail with timeout var MaxWait = 20 * time.Second -//TestEnvironment contains all necessars +// TestEnvironment contains all necessars type TestEnvironment struct { t *testing.T docker *dockertest.Pool @@ -34,7 +34,7 @@ type TestEnvironment struct { Rabbit rabbit } -//NewEnvironment sets up a new environment. It will nlog fatal if something goes wrong +// NewEnvironment sets up a new environment. It will nlog fatal if something goes wrong func NewEnvironment(t *testing.T, dockerTag string) TestEnvironment { t.Helper() tenv := TestEnvironment{t: t} @@ -91,13 +91,13 @@ func (tenv *TestEnvironment) CleanUp() { } } -//ManagementURL returns the full http url including username/password to the management api in the docker environment. +// ManagementURL returns the full http url including username/password to the management api in the docker environment. // e.g. http://guest:guest@localhost:15672 func (tenv *TestEnvironment) ManagementURL() string { return fmt.Sprintf("http://guest:guest@%s:%s", tenv.getHost(), tenv.resource.GetPort("15672/tcp")) } -//AmqpURL returns the url to the rabbitmq server +// AmqpURL returns the url to the rabbitmq server // e.g. amqp://localhost:5672 func (tenv *TestEnvironment) AmqpURL(withCred bool) string { return fmt.Sprintf("amqp://%s:%s", tenv.getHost(), tenv.resource.GetPort("5672/tcp"))