Support for open-telemetry

[kevinrobayna]

I was checking the docs over https://autobase.tech/ and i could not find anywhere how to configure the servers to forward logs or to expose a webserver for the metrics. Is this something you are considering?

If we added support for an open-telemetry agent it would mean anyone would be able to access the information to build their own dashboards.

Also side question regarding automatic maintenance on backups for example, id would be fantastic to allow people to configure heartbeat monitors so that they can be alerted if for some reason pods stop to do this

[vitabaks]

Autobase comes with Netdata pre-installed for system and PostgreSQL monitoring out of the box. However, we fully understand that some teams may prefer to use their own monitoring stack. Since Autobase deploys self-managed database servers under your control, you have full flexibility to install any monitoring tools you prefer — that’s one of the core benefits of a self-hosted DBaaS over traditional managed cloud solutions.

[kevinrobayna]

I see! i didn't notice about netdata, looking at their doc they do not support otel (i was thinking an option would be to locally run also otel and send the data to netdata or whatever you want.)

One issue i see is that if we create a cluster with 3 nodes and then manually change it to support otel then if we add another node to the cluster we need to remember to do the changes there too.

Have you considered ever moving to something like grafana locally? It's equally open-source and we could always run the otel agent and install grafana (and expose it if people want) or just sent the data somewhere else so that we can keep the boxes close down.

[vitabaks]

One issue i see is that if we create a cluster with 3 nodes and then manually change it to support otel then if we add another node to the cluster we need to remember to do the changes there too.

Try pre_deploy_command or post_deploy_command variables https://github.com/vitabaks/autobase/blob/2.2.0/automation/vars/system.yml#L226

Have you considered ever moving to something like grafana locally?

Here is my version of pgwatch with grafana, which I use for my database clusters and which I recommend to my clients, whom I personally support.

https://gitlab.com/postgres-ai/pgwatch2

[kevinrobayna]

Try pre_deploy_command or post_deploy_command variables
https://github.com/vitabaks/autobase/blob/2.2.0/automation/vars/system.yml#L226

oh!! that's awesome!

Here is my version of pgwatch with grafana, which I use for my database clusters and which I recommend to my clients, whom I personally support.

Will def take a look! I'm currently exploring options to deploy a db for a project i'm starting and I wanted to self manage the postgres cluster from the start as i def do not want to do a data migration half way there 🤣