Skip to content

6.2.1 to 6.3.0

Hooks

Hook names have been unified by removing the _hook suffix from the few hooks which used it, e.g. offline_message_hook is now called offline_message. This change affects the hook metric names as well.

Some hooks have been removed:

  • xmpp_stanza_dropped,
  • privacy_list_push,
  • xmpp_bounce_message,
  • mam_flush_messages,
  • mam_muc_flush_messages.

Listener configuration

listen.c2s.c2s_state_timeout option has been renamed to listen.c2s.state_timeout.

Metrics and instrumentation

Metrics implementation has been reworked, and now a Prometheus endpoint is available. For more details, see the instrumentation documentation. This requires some changes to the configuration file, which are described in the following sections.

New [instrumentation] section

To have metrics available in MongooseIM, an [instrumentation] section is now required.

general section

general.all_metrics_are_global has been replaced by instrumentation.exometer.all_metrics_are_global. This option has no effect when using Prometheus, and should be removed from the config file in this case.

Exometer configuration changes

Exometer is no longer configured through app.config, but through mongooseim.toml.

Example configuration that is equivalent to the old example configuration provided in app.config: 

1
2
3
4
5
6
7
[[instrumentation.exometer.report.graphite]]
  host = "127.0.0.1"
  port = 2003
  api_key = ""
  interval = 60000
  prefix = "mongooseim"
  connect_timeout = 5000

See Exometer configuration doc for more information.

New Prometheus endpoint

Alternatively to Exometer, MongooseIM can now expose an endpoint serving Prometheus metrics. The Prometheus metrics provide labels, for example host_type. Because of this, the names of the metrics are usually shorter than the Exometer ones.

The simplest configuration to have it available on port 9090 on the /metrcis path is:

  • Add prometheus handler to the instrumentation section: 
    1
    [instrumentation.prometheus]
  • Configure Prometheus HTTP handler in the listen section: 
    1
2
3
4
5
6
    [[listen.http]]
  port = 9090

  [[listen.http.handlers.mongoose_prometheus_handler]]
    host = "_"
    path = "/metrics"

See instrumentation configuration doc and Prometheus endpoint configuration doc for more information.

Hook metrics

Hook metrics that have been removed:

  • mam_flush_messages,
  • privacy_list_push,
  • xmpp_bounce_message.

All the remaining hook metrics now have the hook_ prefix.

Other metric changes

Several metrics have had their names changed. The new metric names have been documented in module documentation pages.

Upgrade procedure

In order to have metrics available in MongooseIM, perform the following steps. Stop the cluster, or individual nodes, if performing a rolling upgrade, and execute the configuration changes, that were described above:

  1. Add an [instrumentation] section.
  2. Replace the general.all_metrics_are_global option with instrumentation.exometer.all_metrics_are_global.
  3. Configure Exometer exporters in the instrumentation section.
  4. Note that many metrics have new names, and some have been removed.
  1. Add an [instrumentation] section.
  2. Remove the general.all_metrics_are_global option.
  3. Add Prometheus to the instrumentation section.
  4. Configure a listener for Prometheus.

Restart the node or cluster.

Support for the new RDBMS database - CockroachDB

We've introduced support for a new RDBMS database, CockroachDB.

To configure CockroachDB, update the connection settings in your configuration file as follows: 

1
2
3
4
5
6
7
[outgoing_pools.rdbms.default.connection]
  driver = "cockroachdb"
  host = "localhost"
  port = 26257
  database = "mongooseim"
  username = "mongooseim"
  password = "mongooseim_secret"

Refer to the Outgoing connections configuration doc for more information.