Skip to main content

This documentation is intended for internal use by the GDS community.


The Reliability Engineering team uses Logit to provision, manage and ensure availability of your logging infrastructure and provide ELK (Elasticsearch, Logstash, and Kibana) stacks.

The Reliability Engineering team helps you integrate Logit into your environment, and create your logging and usage policies.

Get started with Logit

If you do not have a Logit account you can create one by signing into Logit for the first time. Once you’ve created an account you must always sign into Logit the same way.

To sign into Logit:

  1. Go to
  2. Switch to the “Single Sign-on” tab (note: this is currently very light grey text on a white background, so you may struggle to find it)
  3. Enter your “Company Email Address”
  4. Submit the form to complete Google Authentication.
  5. You’re now signed into the Government Digital Service Dashboard.

Once your user account is created, you’ll need access to your team’s logging stacks. Your team’s Tech Lead should be a Logit administrator who will add your account.

Remove access to Logit

When people leave GDS their G Suite account is cancelled, so they’ll lose access to Logit but the actual Logit account will not be deleted.

It’s important that when people leave GDS, you also delete their logit account so it’s clear who in GDS has access.

Send logs to Logit

Logit offers logstash endpoints. In their default mode there is no authentication on them, this means anyone who has the logstash address could inject fake logs into your stack.

Provision a new stack

After signing into create a new stack by selecting Add stack from the Stacks section of the dashboard.

If you do not see the Add stack button, it means you do not have permission to add a new stack. The tech lead for your team should have permissions to add a new stack.

Contact Reliabilty Engineering using the #reliability-engineering Slack channel for help with permissions.

Stack name

When you choose a stack name, other teams will use this name to identify your stack from their own.

Your team will also need to differentiate between stacks they own. Your stack name should be specific, concise, show who owns it and which environment it serves.

For example:

  • good - GOV.UK Verify Hub Production (Green)
  • bad - New dev environment 2

Stack version

In the Stack version field select the latest Elasticsearch, Logstash and Kibana (ELK) version from the displayed drop-down menu, unless you have specific reasons to prefer an earlier version.

Complete the Daily logs and Retention fields as appropriate then choose continue. Confirm your stack details are correct and select Create Stack.

Once created, return to the Dashboard.

Configure a new stack

Id and Stack API Key

An Id and Stack API Key will have been generated for your new stack. To view this information, choose Settings from the Dashboard.

The Id and Stack API Key will both be UUIDs. We’ll use the UUID below in our examples for the remainder of this manual.


Logstash endpoint

The Logstash endpoint is formed from the stack ID. Our example stack’s Logstash endpoint is

Logstash speaks many different protocols and these are all assigned to sequential TCP ports which are visible from Logstash Inputs under the stack settings page.

TCP ports are assigned on a per-stack basis at stack creation time and will be different per stack.

Send logs from PaaS to logit

You can send logs from PaaS to logit

Ship logs using Amazon S3

In some cases such as logging for Amazon services (like, CloudTrail, VPC FlowLogs) it can make sense to use an S3 bucket as a log “source” for your Logit stack. This differs from other ways of using Logit as logs are “pulled” into your stack by a process in the Logit infrastructure.

Share access to resources in your Amazon account

Amazon recommend using IAM Roles to share access to resources in your account. This method doesn’t require the creation of shared secrets.

Currently Logit don’t support this, however they do plan to support it in the future. Right now the only way to grant them access is by issuing them access keys which can access the S3 bucket in question.

When setting this up you make sure that

  • The access key is only able to ListObjects and GetObject from the S3 bucket containing the logs
  • You are able to migrate to Role based access in the future when it is supported by Logit

Configure Logit

Currently pulling logs from an S3 bucket requires a Logit support engineer to configure it for your stack. To start this process, add a “Source” to your Logit stack and select “S3”. Complete the form and a Logit support engineer will contact you once it has been configured.

Logs pulled from S3 will be processed as usual by Logstash.

Shut down unsecure inputs

Logit stacks have a non-restrictive security configuration by default. Each of the inputs listed will be behind a TCP port that is open to the internet at large, and will not authenticate connecting clients.

The only inputs we are interested in are Beats-SSL and Syslog-SSL, you will need to log a support request with Logit to close the TCP or UDP ports for all other protocols.

If you know in advance which IP addresses or address ranges you will be sourcing logging traffic from, include these in your request so that the Beats-SSL and Syslog-SSL ports can be safelisted.

Configure Filebeat

Filebeat is a Logstash client that reads log files from the filesystem and ships them to Logstash as they are being written to. Please refer to the Filebeat documentation for details on installation, general configuration and operation.

The following is a sample configuration fragment for shipping to Logit. The port number in the hosts specification must be the TCP port assigned to the Beats-SSL input in your stack.

  enabled: true
    enabled: true

Configure Rsyslog

Rsyslog version 8.x is known to work, we have not tried earlier versions. Although Ubuntu Trusty ships with 7.x, packages of 8.x for Trusty are available on the Rsyslog vendor’s PPA. You must also install the optional rsyslog-gnutls package.

Then, load the gtls “stream driver” and configure it for operation with Logit:

$DefaultNetstreamDriver gtls
$ActionSendStreamDriverAuthMode x509/name
$ActionSendStreamDriverMode 1

Declare an action on matching messages that uses the omfwd module to forward them to Logit:

*.*;local2.none action (

Note the streamdriver="gtls" and streamdrivermode="1" are required here, despite the earlier config fragment implying that the default behaviour has been changed.

Proxy patterns

Architectural constraints around egress security often mean that having each of your servers connect directly to Logit’s endpoint is not an option.

This section covers some possible approaches to securely relaying Beats and Syslog traffic to Logit at your environment’s boundary.

Proxy Beats-SSL (Lumberjack protocol)

The Lumberjack protocol is not HTTP based, and as such, you can’t use an HTTP proxy. Filebeat does, however, support SOCKS negotiation.

In the following example, is the private network CIDR range of the log-generating servers that you’re proxying for, and 12345 is the TCP port of the Beats-SSL input on the Logstash endpoint.

Sample, minimal configuration for a Dante socks proxy:

logoutput: syslog

internal: eth0 port = 1080
external: eth0

method: none
clientmethod: none

user.privileged: proxy
user.notprivileged: nobody
user.libwrap: nobody

client pass {
  method: none
  log: connect disconnect error

client block {
  log: connect error

pass {
  from: to: port = 12345
  protocol: tcp
  command: connect
  log: connect disconnect error

Proxy Syslog

This section is under review.

The Syslog protocol is designed to be relayed; Rsyslog fully supports this.

The configuration example in the above section will already do the egress-proxy-to-Logit part of the relay unmodified, and this should be applied to your egress proxy’s rsyslogd config.

However, you may need some additional configuration to allow your egress proxy’s rsyslogd to accept syslog messages from your internal hosts. For example:

$ModLoad imudp
$UDPServerRun 514

Logit incident management

To find out about technical incidents affecting Logit you must:

Reliability Engineering uses these channels to tell GDS Logit users:

  • about incidents such as Kibana or ElasticSearch being unavailable
  • when an incident is resolved and normal service is resumed

Logit will keep in contact with Reliability Engineering during an incident, and ensure their Logit status page is up to date.

Report an incident

If you discover Logit or Kibana have failed, the process you must follow will differ depending on whether you are working during office hours or out-of-hours.

During office hours (9.30am to 5.30pm)

  1. Check the Logit status page to see if Logit know about the incident.
  2. Join the #reliability-eng Slack channel and describe the problem.
  3. Get confirmation that a Reliability Engineering team member has taken over the incident.
  4. Wait for updates from Reliability Engineering announcements or the #reliability-eng Slack channel.

The Reliability Engineering Tools team will work directly with Logit coordinating communication on behalf of GDS.

Outside office hours (from 5.30pm until 9.30am)

Logit provide an automated support process for out of hours incidents, for example in the event of a platform-wide outage an on-call Logit engineer is automaticlly paged. When you contact Logit outside office hours you need the Logit support number and PIN.

How to report a Logit incident out-of-hours:

  1. Use the Logit status page to check if Logit know about the incident. If Logit have updated their status page, check this for updates on a half-hourly basis until the incident is resolved.
  2. If Logit are unaware of the issue, and it’s urgent, you can telephone Logit to wake an on-call engineer.
  3. Call the Logit support number and enter the pin to leave your message.
  4. Describe the incident leaving your name, and contact number, Logit will call you back within 30 minutes to acknowledge the incident.

Logit provides updates:

  • directly to you if the incident is stack related
  • on the Logit status page if platform related
  1. Join the #reliability-eng Slack channel and describe the issue so Reliability Engineering can contact you when office hours resume.
  2. When office hours resume, a Reliability Engineering team member will confirm they have taken over the incident.
  3. Wait for updates from Reliability Engineering announcements or the #reliability-eng Slack channel.

The Reliability Engineering Tools team will work directly with Logit coordinating communication on behalf of GDS.

Incident reviews

If Logit have a major incident, for example a full Logit outage, Reliability Engineering will announce an internal incident review using the Reliability Engineering announcements Google Group explaining how to register your interest in attending.

After the incident review meeting, Reliability Engineering will:

  1. Email the incident report including a confirmed set of recommendations to Reliability Engineering announcements.
  2. Follow up on actions arising from the incident review.
  3. Follow up with Logit on any actions on their side, including claiming refunds for the month in question according to our Enterprise SLA which is 99.9% uptime.
  4. Email Logit’s own incident report to Reliability Engineering announcements, Logit’s incident report will be sent out within 14 days of the incident.


Contact Reliabilty Engineering using the #reliability-eng Slack channel.