Self-hosted disk space usage

So with the self-hosted version, I’ve got around 800GB of storage that I’ve allocated to my node.

What I’m currently wondering is what happens when the node storage is full?

Do workspaces that have been deleted eventually get the space reclaimed?

good question, i wonder if there is a way to limit the number of workspaces a user can have

The answer to my original question is it performs housekeeping at an unknown interval. I left it for a few days as I got busy and noticed storage dropped.

As for limiting workspaces, I don’t think so. What can be done instead is limiting the resources available to each one.

We recently shipped support per-workspace disk space quotas at

I would alaso be interested in a definitive answer to this…

I have a setup (0.10.0 self-hosted) where one of the worker nodes gets filled up quite quickly, even though we have 300GB disk space allocated to /mnt/persistent which is where the workspaces and other data are.

I have had the node run into problems with disk space and have been manually deleting what I thought was safe to delete from the workspaces directory.

I would prefer to be able to set and foget this - or trigger a clean up in some way.

1 Like

I agree better documentation related to disk space management is required. Raised a feedback item over at Administration of Gitpod Self-Hosted by mashing the :frowning: button at the bottom. Internally we have an ops handbook for these scenarios and have started discussions about opening up the knowledge. Internal link.

When do these new features get pushed to self hosted?

Gitpod should follow the gitlab idea

Two versions

Gitpod-ce and gitpob-ee

So until this is made available, any useful clues on things that can be done when it comes to disk space management?

I’d like to be able to be proactive rather than running into issues with nodes’ diskspace filling up because old workspaces are not being clearer.

I started out with an 800GB HDD partition for storing the workspaces but it was extremely slow, especially when the actual repos I was pulling were stored on the same host disk. I switched to a 500GB NVMe partition and it made a world of difference. The other thing I did to help with that was making workspace resources limited with defined limits that can be modified to suit your needs listed here