leftwo
commented
2 weeks ago In dsc/src/main.rs, we have the following code that starts a downstairs:
fn start(&self) -> Result<Child> {
println!("Make output file at {:?}", self.output_file);
let outputs = File::create(&self.output_file)
.context("Failed to create output file")?;
let errors = outputs.try_clone()?;
let port_value = format!("{}", self.port);
let mode = if self.read_only {
"ro".to_string()
} else {
"rw".to_string()
};
let region_dir = self.region_dir.clone();
let cmd = Command::new(self.ds_bin.clone())
.args([
"run",
"-p",
&port_value,
"-d",
®ion_dir,
"--mode",
&mode,
])
.stdout(Stdio::from(outputs))
.stderr(Stdio::from(errors))
.spawn()
.context("Failed trying to run downstairs")?;
println!(
"Downstairs {} port {} PID:{:?}",
region_dir,
self.port,
cmd.id()
); In our logs above, we see the first message Make outfile at... for this client, but we never see the 2nd, Downstairs {} port {}.... So, that command (which does spawn, we see the output file) never comes back from that spawn, as near as I can determine.
mkeeter
In CI (link may be bad) job: https://buildomat.eng.oxide.computer/wg/0/artefact/01J9KZDBT7Q76BMAZ3NFT2EM6G/JHuMk44VV03fuovquSYDTdL9uCEruARhbYRD7WhRUTif7Lam/01J9KZDV6QXGVBQG0ZHA6JNXTH/01J9M3TRMJHBD7KTGG9QWWK022/dsc-out.txt
We had the test timeout. Looking at test log output, we can see a new check we added is waiting for all the downstairs to respond:
The waiting for it waits till the test hits a timeout and we abort it. In the logs for
dscwe can see it got toStartingon client 0, but never toRunning:A third bit of information, if I look at the output for client 0, port 8810, I do see that it started:
This suggests the problem is somewhere in the messages between different parts of
dsc.