mirror of
https://github.com/neondatabase/neon.git
synced 2026-01-08 14:02:55 +00:00
ba498a630a
Part of https://github.com/neondatabase/cloud/issues/13127. Resolves #9153 What changed in this PR: 1. Adds `ComputeSpec.disk_quota_bytes: Option<u64>` 2. Adds new arg to compute_ctl: `--set-disk-quota-for-fs <mountpoint>` 3. Implements running `/neonvm/bin/set-disk-quota` with the right value if both cmdline arg AND field in the spec are specified 4. Patches `/etc/sudoers.d` to allow `compute_ctl` to set quota with sudo This PR is very similar to the swap support added earlier, you can take a look at it as prior art: #7434 In theory, it can be implemented outside of compute_ctl when we will have a separate neonvm daemon, but we are not there yet. Current implementation is the simplest possible to unblock computes with larger disks. All code related to usage of `/neonvm/bin/set-disk-quota` is located in `disk_quota.rs`. We need to call this script with the following arguments: `/neonvm/bin/set-disk-quota {size_kb} {mountpoint}`. Quotas are set on the filesystem level, so we need to provide path to the directory that filesystem was mounted to. I tested this change locally with https://github.com/neondatabase/cloud/pull/17270. It should be safe to merge, because this feature is gated by both cmdline arg and field in the spec. If control-plane doesn't set values in both places, compute_ctl won't be affected by this change.