ML7 I know to have been reinstalled and it lost its sonar install.
ML4 is a little more curious but it hasn't been reporting since nov 11. It's been up for longer than that. Sonar appears to run. It tries to send mail about something to me, but the mail never arrives, so mail is misconfigured too (too many things don't work about email to be able to rely on it at all, see #585).
Looks like ML4 could be affected by the problem with parsing rocm-smi output:
[larstha@ml4 sonar-main]$ git checkout release_0_12
branch 'release_0_12' set up to track 'origin/release_0_12'.
Switched to a new branch 'release_0_12'
[larstha@ml4 sonar-main]$ cargo build
Compiling sonar v0.12.2 (/itf-fi-ml/home/larstha/p/sonar-main)
Finished dev [unoptimized + debuginfo] target(s) in 1.87s
[larstha@ml4 sonar-main]$ target/debug/sonar ps
thread 'main' panicked at src/amd.rs:223:22:
called `Option::unwrap()` on a `None` value
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
and then
[larstha@ml4 sonar-main]$ git checkout main
Switched to branch 'main'
Your branch is up to date with 'origin/main'.
[larstha@ml4 sonar-main]$ cargo build
Compiling sonar v0.13.0-devel (/itf-fi-ml/home/larstha/p/sonar-main)
Finished dev [unoptimized + debuginfo] target(s) in 1.75s
[larstha@ml4 sonar-main]$ target/debug/sonar ps
v=0.13.0-devel,time=2024-11-19T09:04:28+01:00,host=ml4.hpc.uio.no,user=root,cmd=xfsalloc,pid=1376,ppid=2
v=0.13.0-devel,time=2024-11-19T09:04:28+01:00,host=ml4.hpc.uio.no,user=root,cmd=ksoftirqd/60,pid=383,ppid=2,cputime_sec=3
v=0.13.0-devel,time=2024-11-19T09:04:28+01:00,host=ml4.hpc.uio.no,user=root,cmd=ksoftirqd/8,pid=65,ppid=2,cputime_sec=1
...
So the AMD parsing thing (probably) should be backported to v0.12 or we should move to v0.13-devel on ML4.
ML7 I know to have been reinstalled and it lost its sonar install.
ML4 is a little more curious but it hasn't been reporting since nov 11. It's been up for longer than that. Sonar appears to run. It tries to send mail about something to me, but the mail never arrives, so mail is misconfigured too (too many things don't work about email to be able to rely on it at all, see #585).
Looks like ML4 could be affected by the problem with parsing rocm-smi output:
and then
So the AMD parsing thing (probably) should be backported to v0.12 or we should move to v0.13-devel on ML4.