PostgreSQL Add-ons
Overview
Our PostgreSQL deployments support the following add-ons:
Syslog-NG
Configuration instructions for the syslog-ng add-on can be found on the Syslog-NG Setup page.
Logs are forwarded continuously from all the capsules present on your deployment; so you will see database logs and haproxy logs in chronological order. For easier parsing, Compose adds a header to each log message that contains a timestamp, hostname, and process information.
As a result, the raw messages that arrive at your service will vary.
Papertrail
Raw Message:
<timestamp> <hostname>: <binary[pid]>: <type info> | <haproxy status, or PostgreSQL log>
Loggly
Raw Message:
<hostname> : [<customer [email protected]>] <binary[pid]>: <type info> | <haproxy status or PostgreSQL log>
NB: In addition to just displaying the raw information from syslog-ng, Loggly also applies some parsing by default. It does not recognize the format of our logs, so you may want to implement your own parsing or derived fields to make more effective use of the logs there.
Logentries
Logentries is no longer available for new deployments or as a new add-on for existing deployments. The format is here for deployments where it is already configured.
Raw Message:
<timestamp> <hostname>: <binary[pid]>: <type info> | <haproxy status, or PostgreSQL log>
Telegraf
Telegraf for PostgreSQL uses the statistics from the PostgreSQL's own pg_stat_database and pg_stat_bgwriter views. Which stats are available depend on PostgreSQL version and are listed in the telegraf PostgreSQL plugin README.
Installing the telegraf add-on to your Compose deployment allows the statistics to be forwarded to either metrics service: Datadog or Librato.
Configuration instructions for the Telegraf add-on can be found on the Telegraf page.
Still Need Help?
If this article didn't solve things, summon a human and get some help
Updated over 3 years ago