-
Notifications
You must be signed in to change notification settings - Fork 6
/
slurm-job-exporter-el7.spec
82 lines (72 loc) · 3.8 KB
/
slurm-job-exporter-el7.spec
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
Name: slurm-job-exporter
Version: 0.4.2
Release: 1%{?dist}
Summary: Prometheus exporter for stats in slurm accounting cgroups
License: Apache License 2.0
URL: https://github.com/guilbaults/slurm-job-exporter
Source0: https://github.com/guilbaults/%{name}/archive/refs/tags/v%{version}.tar.gz
BuildArch: noarch
BuildRequires: systemd
Requires: python3
Requires: python36-psutil
%description
Prometheus exporter for the stats in the cgroup accounting with slurm. This will also collect stats of a job using NVIDIA GPUs.
%prep
%setup -q
%build
%install
mkdir -p %{buildroot}/%{_bindir}
mkdir -p %{buildroot}/%{_unitdir}
sed -i -e '1i#!/usr/bin/python3' slurm-job-exporter.py
install -m 0755 %{name}.py %{buildroot}/%{_bindir}/%{name}
install -m 0744 get_gpus.sh %{buildroot}/%{_bindir}/get_gpus.sh
install -m 0644 slurm-job-exporter.service %{buildroot}/%{_unitdir}/slurm-job-exporter.service
%clean
rm -rf $RPM_BUILD_ROOT
%files
%{_bindir}/%{name}
%{_bindir}/get_gpus.sh
%{_unitdir}/slurm-job-exporter.service
%changelog
* Wed Apr 17 2024 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.4.2-1
- Adding support for L4 that lack FP64
* Wed Mar 27 2024 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.4.1-1
- Handle when no GPU are allocated in a job and improve systemd service
* Mon Jan 22 2024 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.4.0-1
- Adding support for cgroup v2
* Fri Nov 24 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.3.2-1
- Fixing support for GPUs without FP64
* Tue Nov 23 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.3.1-1
- Adding support for GPUs without FP64
* Tue Sep 25 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.3.0-1
- Removing the dependency on cgexec
* Tue Aug 1 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.2.0-1
- Adding counter_process_usage to measure and name the processes inside the cgroups
* Fri Jul 14 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.1.0-1
- Adding support for MIG devices
* Tue Apr 11 2023 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.13-1
- Fixing DCGM bandwidth gauges and adding --dcgm-update-interval parameter
* Mon Nov 29 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.12-1
- slurm_job_threads_count by thread status, can now distinguish sleeping, running, io-wait ...
* Mon Nov 28 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.11-1
- Measure the qty of processes and threads
* Mon Oct 31 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.10-1
- Adding support of GPU-less job on a GPU node
* Tue Oct 18 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.9-1
- Power stats is now optional since its not visible in NVIDIA-GRID
* Fri Jul 15 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.8-1
- Handle utf8 characters in job's environment
* Tue Jul 5 2022 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.7-1
- Collecting GPU stats with Nvidia DCGM and keeping NVML as a fallback
* Fri Aug 13 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.6-1
- Collecting more memory stats from each job cgroup
* Wed Apr 28 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.5-1
- Adding a workaround to prevent a crash when the other nodes of a multinodes job is using slurm_adopt
* Wed Apr 14 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.4-1
- Grabbing slurm account and using a better way to match GPU to a job
* Wed Apr 7 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.3-1
- Replacing job by slurmjobid since it conflit with a internal field in Prometheus
* Wed Mar 10 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.2-1
- Fixing a collecting bug, removing the sleep() and changing the metric prefixes to slurm_job
* Mon Mar 8 2021 Simon Guilbault <simon.guilbault@calculquebec.ca> 0.0.1-1
- Initial release