Skip to main content
This page describes what to do in case of an Icinga alert. For more information you could search the govuk-puppet repo for the source of the alert
Last updated: 22 Jun 2021

App healthcheck not ok

Many apps on GOV.UK have a healthcheck endpoint.

  • This is usually /healthcheck [1].
  • Some apps just pick a random page [1].

The alert works by making a request for the healthcheck endpoint on the machine where the app is running.

Check if the app is running

If the alert is appearing alongside an upstart not up alert, it’s likely that the process isn’t even running, therefore any requests to the healthcheck endpoint will also fail.

Test the healthcheck endpoint manually

You can SSH onto the machine and curl it yourself.

# SSH on to a machine running Content Publisher
gds govuk connect ssh -e integration backend

# Find the port it's running on
ps -ef | grep content-publisher | grep master

# Do the Icinga check manually
curl localhost:3221/healthcheck

Apps with a custom healthcheck endpoint often make use of the generic checks in govuk_app_config. Some apps also implement custom checks, and the alert should link to custom documentation to explain these.

Connection Refused Error

This means the app is not accepting requests for the healthcheck endpoint, and is probably down.

Timeout Error

This means the app is accepting requests, but taking too long to respond (over 20 seconds).

  • Try the healthcheck endpoint manually, as above, to confirm.
  • Check the logs e.g. tail -100f /var/log/email-alert-api/app.err.log.
  • Check for resource issues e.g. on the Machine dashboard.

Active Record Check

This means the app is unable to connect to its database.

# SSH on to a machine with the problem
gds govuk connect ssh -e integration backend

# Find the DB connection details
govuk_setenv content-publisher env | grep -i postgres

# Try to connect
psql -h postgresql-primary -U content-publisher -W content_publisher_production

# Try a command
select * from users;

Sidekiq / Redis Check

This means that the Sidekiq workers can’t connect to Redis.

  • Check for any Redis alerts.
  • Check for network connectivity to Redis.
# SSH on to a machine with the problem
gds govuk connect ssh -e integration backend

# Find the redis connection details
govuk_setenv content-publisher env | grep -i redis

# Try to connect
redis-cli -h backend-redis

# Try a command
keys *

Sidekiq Retry Size Check

This means that Sidekiq jobs are failing.

  • Check the Sidekiq ‘Retry set size’ graph to see if we have a high number of failed jobs.
  • Are the workers reporting any problems or any issues being raised in Sentry?
  • Check Kibana for Sidekiq error logs (application: <app> AND @type: sidekiq).

Sidekiq Queue Latency Check

This means the time it takes for a Sidekiq job to be processed is unusually high.