If you're the week's support hero or you are providing support for a customer and they have questions about their self-hosted deployment, follow this guide to provide initial support before looping in someone from the Infrastructure team.
Gather basic information
Here's a sample message that should help gather the relevant information up-front (appropriate for #community-support
, but if working in a private channel with a paid customer, remove some of the obvious questions).
👋 Are you self-hosting or on PostHog Cloud? (if self-hosting please answer below)
- What have you tried to solve the issue so far? How did that go?
- Which cloud provider are you using? How many nodes are you running?
- Are you using our Helm chart to deploy PostHog? Have you make any customisations? Can you please share your
values.yaml
file?- If you have any pod(s) erroring/restarting, can you please share the logs?
- Do you have any kind of monitoring configured? (if not, can you please enable at least Grafana and Prometheus in the
values.yaml
of the Helm chart?)- How many events are in ClickHouse, and how many were ingested last month?
Kickstart the debugging process
What's the output of
kubectl get pods -n posthog
?This should look something like this:
TerminalNAME READY STATUS RESTARTS AGEchi-posthog-posthog-0-0-0 1/1 Running 0 11dclickhouse-operator-6b5438eh5fb-bt5fk 2/2 Running 0 11dposthog-beat-7782927b778-wxvhl 1/1 Running 0 11dposthog-cert-manager-69fahs7b57-c48dn 1/1 Running 0 11dposthog-cert-manager-cainjector-6d95d93mn8-6tz6k 1/1 Running 0 11dposthog-cert-manager-webhook-6469930mdfc-6l958 1/1 Running 0 11dposthog-events-55283995cc-rpjdm 1/1 Running 0 11dposthog-ingress-nginx-controller-648bdn892f-w7qhp 1/1 Running 0 11dposthog-pgbouncer-77fb4djs85c-2d24t 1/1 Running 0 11dposthog-plugins-54fjd8292649-66gsm 1/1 Running 0 18mposthog-posthog-kafka-0 1/1 Running 0 11dposthog-posthog-postgresql-0 1/1 Running 0 11dposthog-posthog-redis-master-0 1/1 Running 0 11dposthog-posthog-zookeeper-0 1/1 Running 0 11dposthog-posthog-zookeeper-1 1/1 Running 0 11dposthog-posthog-zookeeper-2 1/1 Running 0 11dposthog-web-78dns2f5d7-6zdkc 1/1 Running 0 11dposthog-worker-7857nd8268-j8c4f 1/1 Running 0 11dWhen they send you the output from the command in step 1, if any of the pods has a status other than
Running
, ask them for the output ofkubectl logs pod-name -n posthog
The output from the previous step may or may not be familiar to you. Sometimes the logs will be something you've seen before. If that's the case, try to reproduce the issue locally and come up with a fix. If things are cryptic to you, loop in someone from the Infrastructure team.
If a pod is listed as
Failed
, suggest that they try an upgrade withhelm upgrade -f values.yaml -n posthog
Common issues
Some common issues you might encounter are:
PostHog is stuck migrating/the migrate job has an issue
Tell them to run the following:
kubectl -n posthog delete job $(kubectl -n posthog get jobs --no-headers -o custom-columns=NAME:.metadata.name)helm upgrade -f values.yaml -n posthog
The app/plugin server has an issue
The first thing that you can safely try here is to tell them to restart the apps pod:
# terminate the running plugins podkubectl scale deployment posthog-plugins --replicas=0 -n posthog# start a new plugins podkubectl scale deployment posthog-plugins --replicas=1 -n posthog
How can I connect to Postgres?
Send them a link to this doc.
Kafka crashed / out of storage
Send them a link to this doc.
Connection is not secure / DNS problems
Before looping in someone, ask them to check that DNS records are correctly set and have propagated with this command:
nslookup <your-hostname> 1.1.1.1
Other issues
Check out our Troubleshooting page for other common issues and how you might be able to provide "first aid" before looping in someone.
All is lost
The idea of this doc is to cover some basic support that you can provide in order to either help the customer solve their issue or gather basic info before someone from the Infrastructure team shows up.
However, never hesitate to call us! We're more than happy to help.
Also, if things seem very serious and/or relate to a paying customer, do reach out to us right away.