rust/neon
1
0
Fork 0
mirror of https://github.com/neondatabase/neon.git synced 2026-01-04 12:02:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

compute_ctl: start pg in cgroup for vms (#4920)

Browse Source 
Starts `postgres` in cgroup directly from `compute_ctl` instead of from
`vm-builder`. This is required because the `vm-monitor` cannot be in the
cgroup it is managing. Otherwise, it itself would be frozen when
freezing the cgroup.

Requires https://github.com/neondatabase/cloud/pull/6331, which adds the
`AUTOSCALING` environment variable letting `compute_ctl` know to start
`postgres` in the cgroup.

Requires https://github.com/neondatabase/autoscaling/pull/468, which
prevents `vm-builder` from starting the monitor and putting postgres in
a cgroup. This will require a `VM_BUILDER_VERSION` bump.
This commit is contained in:
Felix Prasanna
2023-08-25 15:59:12 -04:00
committed by GitHub
parent 40268dcd8d
commit 7b5489a0bb
2 changed files with 27 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
.github/workflows/build_and_test.yml vendored
View File

@@ -752,7 +752,7 @@ jobs:
run:
shell: sh -eu {0}
env:
VM_BUILDER_VERSION: v0.16.3
VM_BUILDER_VERSION: v0.17.5
steps:
- name: Checkout
@@ -775,6 +775,7 @@ jobs:
run: |
./vm-builder \
-enable-file-cache \
-cgroup-uid=postgres \
-src=369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}} \
-dst=369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}

28
compute_tools/src/compute.rs
View File

@@ -1,4 +1,5 @@
use std::collections::HashMap;
use std::env;
use std::fs;
use std::io::BufRead;
use std::os::unix::fs::PermissionsExt;
@@ -175,6 +176,27 @@ impl TryFrom<ComputeSpec> for ParsedSpec {
}
}
/// If we are a VM, returns a [`Command`] that will run in the `neon-postgres`
/// cgroup. Otherwise returns the default `Command::new(cmd)`
///
/// This function should be used to start postgres, as it will start it in the
/// neon-postgres cgroup if we are a VM. This allows autoscaling to control
/// postgres' resource usage. The cgroup will exist in VMs because vm-builder
/// creates it during the sysinit phase of its inittab.
fn maybe_cgexec(cmd: &str) -> Command {
// The cplane sets this env var for autoscaling computes.
// use `var_os` so we don't have to worry about the variable being valid
// unicode. Should never be an concern . . . but just in case
if env::var_os("AUTOSCALING").is_some() {
let mut command = Command::new("cgexec");
command.args(["-g", "memory:neon-postgres"]);
command.arg(cmd);
command
} else {
Command::new(cmd)
}
}
/// Create special neon_superuser role, that's a slightly nerfed version of a real superuser
/// that we give to customers
fn create_neon_superuser(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
@@ -451,7 +473,7 @@ impl ComputeNode {
pub fn sync_safekeepers(&self, storage_auth_token: Option<String>) -> Result<Lsn> {
let start_time = Utc::now();
let sync_handle = Command::new(&self.pgbin)
let sync_handle = maybe_cgexec(&self.pgbin)
.args(["--sync-safekeepers"])
.env("PGDATA", &self.pgdata) // we cannot use -D in this mode
.envs(if let Some(storage_auth_token) = &storage_auth_token {
@@ -586,7 +608,7 @@ impl ComputeNode {
// Start postgres
info!("starting postgres");
let mut pg = Command::new(&self.pgbin)
let mut pg = maybe_cgexec(&self.pgbin)
.args(["-D", pgdata])
.spawn()
.expect("cannot start postgres process");
@@ -614,7 +636,7 @@ impl ComputeNode {
let pgdata_path = Path::new(&self.pgdata);
// Run postgres as a child process.
let mut pg = Command::new(&self.pgbin)
let mut pg = maybe_cgexec(&self.pgbin)
.args(["-D", &self.pgdata])
.envs(if let Some(storage_auth_token) = &storage_auth_token {
vec![("NEON_AUTH_TOKEN", storage_auth_token)]