pezkuwichain/pezkuwi-subxt
1
0
Fork 0
mirror of https://github.com/pezkuwichain/pezkuwi-subxt.git synced 2026-06-12 15:51:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Check spawned worker version vs node version before PVF preparation (#6861)

Browse Source 
* Check spawned worker version vs node version before PVF preparation

* Address discussions

* Propagate errors and shutdown preparation and execution pipelines properly

* Add logs; Fix execution worker checks

* Revert "Propagate errors and shutdown preparation and execution pipelines properly"

This reverts commit b96cc3160ff58db5ff001d8ca0bfea9bd4bdd0f2.

* Don't try to shut down; report the condition and exit worker

* Get rid of `VersionMismatch` preparation error

* Merge master

* Add docs; Fix tests

* Update Cargo.lock

* Kill again, but only the main node process

* Move unsafe code to a common safe function

* Fix libc dependency error on MacOS

* pvf spawning: Add some logging, add a small integration test

* Minor fixes

* Restart CI

---------

Co-authored-by: Marcin S <marcin@realemail.net>
This commit is contained in:
s0me0ne-unkn0wn
2023-03-30 00:48:41 +02:00
committed by GitHub
parent 8b1645faf0
commit 55b4aceb99
12 changed files with 156 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
polkadot/node/core/pvf/src/execute/worker.rs
+26 -8
View File
@@ -47,9 +47,13 @@ pub async fn spawn(
executor_params: ExecutorParams,
spawn_timeout: Duration,
) -> Result<(IdleWorker, WorkerHandle), SpawnErr> {
let (mut idle_worker, worker_handle) =
spawn_with_program_path("execute", program_path, &["execute-worker"], spawn_timeout)
.await?;
let (mut idle_worker, worker_handle) = spawn_with_program_path(
"execute",
program_path,
&["execute-worker", "--node-impl-version", env!("SUBSTRATE_CLI_IMPL_VERSION")],
spawn_timeout,
)
.await?;
send_handshake(&mut idle_worker.stream, Handshake { executor_params })
.await
.map_err(|error| {
@@ -260,11 +264,25 @@ impl Response {
}
/// The entrypoint that the spawned execute worker should start with. The `socket_path` specifies
/// the path to the socket used to communicate with the host.
pub fn worker_entrypoint(socket_path: &str) {
/// the path to the socket used to communicate with the host. The `node_version`, if `Some`,
/// is checked against the worker version. A mismatch results in immediate worker termination.
/// `None` is used for tests and in other situations when version check is not necessary.
pub fn worker_entrypoint(socket_path: &str, node_version: Option<&str>) {
worker_event_loop("execute", socket_path, |rt_handle, mut stream| async move {
let handshake = recv_handshake(&mut stream).await?;
let worker_pid = std::process::id();
if let Some(version) = node_version {
if version != env!("SUBSTRATE_CLI_IMPL_VERSION") {
gum::error!(
target: LOG_TARGET,
%worker_pid,
"Node and worker version mismatch, node needs restarting, forcing shutdown",
);
crate::kill_parent_node_in_emergency();
return Err(io::Error::new(io::ErrorKind::Unsupported, "Version mismatch"))
}
}
let handshake = recv_handshake(&mut stream).await?;
let executor = Arc::new(Executor::new(handshake.executor_params).map_err(|e| {
io::Error::new(io::ErrorKind::Other, format!("cannot create executor: {}", e))
})?);
@@ -273,7 +291,7 @@ pub fn worker_entrypoint(socket_path: &str) {
let (artifact_path, params, execution_timeout) = recv_request(&mut stream).await?;
gum::debug!(
target: LOG_TARGET,
worker_pid = %std::process::id(),
%worker_pid,
"worker: validating artifact {}",
artifact_path.display(),
);
@@ -307,7 +325,7 @@ pub fn worker_entrypoint(socket_path: &str) {
// Log if we exceed the timeout and the other thread hasn't finished.
gum::warn!(
target: LOG_TARGET,
worker_pid = %std::process::id(),
%worker_pid,
"execute job took {}ms cpu time, exceeded execute timeout {}ms",
cpu_time_elapsed.as_millis(),
execution_timeout.as_millis(),