Thicket data repository for the EEG
anil.recoil.org
1{
2 "id": "https://www.tunbury.org/2025/07/29/slurm-versions",
3 "title": "Slurm Versions",
4 "link": "https://www.tunbury.org/2025/07/29/slurm-versions/",
5 "updated": "2025-07-29T00:00:00",
6 "published": "2025-07-29T00:00:00",
7 "summary": "Slurm requires both the client and server to be on the same version.",
8 "content": "<p>Slurm requires both the client and server to be on the same version.</p>\n\n<div><div><pre><code>[2025-07-29T15:41:34.492] error: slurm_unpack_received_msg: [[foo.cl.cam.ac.uk]:34214] Invalid Protocol Version 10752 from uid=0: No error\n[2025-07-29T15:41:34.492] error: slurm_unpack_received_msg: [[foo.cl.cam.ac.uk]:34214] Incompatible versions of client and server code\n[2025-07-29T15:41:34.502] error: slurm_receive_msg [128.232.93.254:34214]: Incompatible versions of client and server code\n</code></pre></div></div>\n\n<p>Noble (24.04) has Slurm 23.11.4-1.2ubuntu5, whereas Plucky (25.04) has 24.11.3-2.</p>\n\n<p>The latest version is 25.05.1. <a href=\"https://www.schedmd.com/download-slurm\">https://www.schedmd.com/download-slurm</a>.</p>\n\n<p>The recommended approach is to build the Debian <code>.deb</code> packages from source. First, install basic Debian package build requirements:</p>\n\n<div><div><pre><code>apt <span>install </span>build-essential fakeroot devscripts equivs\n</code></pre></div></div>\n\n<p>Unpack the distributed tarball:</p>\n<div><div><pre><code>curl <span>-L</span> https://download.schedmd.com/slurm/slurm-25.05.1.tar.bz2 | <span>tar</span> <span>-xajf</span> - <span>&&</span> <span>cd </span>slurm-25.05.1\n</code></pre></div></div>\n\n<p>Install the Slurm package dependencies:</p>\n<div><div><pre><code>mk-build-deps <span>-i</span> debian/control\n</code></pre></div></div>\n\n<p>Build the Slurm packages:</p>\n<div><div><pre><code>debuild <span>-b</span> <span>-uc</span> <span>-us</span>\n</code></pre></div></div>\n\n<blockquote>\n <p>Before installing, ensure any old installations have been removed with <code>apt remove slurm*</code> and <code>apt remove libslurm*</code>.</p>\n</blockquote>\n\n<h1>Worker</h1>\n\n<div><div><pre><code>dpkg <span>-i</span> slurm-smd-slurmd_25.05.1-1_amd64.deb slurm-smd-client_25.05.1-1_amd64.deb slurm-smd_25.05.1-1_amd64.deb slurm-smd_25.05.1-1_amd64.deb\n</code></pre></div></div>\n\n<h1>Head controller</h1>\n\n<div><div><pre><code>dpkg <span>-i</span> slurm-smd-slurmctld_25.05.1-1_amd64.deb slurm-smd-client_25.05.1-1_amd64.deb slurm-smd_25.05.1-1_amd64.deb slurm-smd_25.05.1-1_amd64.deb\n</code></pre></div></div>\n\n<p>With the same version of Slurm on both machines, the instructions from my earlier <a href=\"https://www.tunbury.org/2025/04/14/slurm-workload-manager/\">post</a> are working again.</p>\n\n<div><div><pre><code><span># sinfo</span>\nPARTITION AVAIL TIMELIMIT NODES STATE NODELIST\neeg<span>*</span> up infinite 1 idle foo\n\n<span># srun -N1 -l /bin/hostname</span>\n0: foo.cl.cam.ac.uk\n</code></pre></div></div>\n\n<p>Slurm communicates directly over TCP connections using ports 6817/6818, so ensure that no firewalls are in the way!</p>",
9 "content_type": "html",
10 "author": {
11 "name": "Mark Elvers",
12 "email": "mark.elvers@tunbury.org",
13 "uri": null
14 },
15 "categories": [
16 "Slurm",
17 "tunbury.org"
18 ],
19 "source": "https://www.tunbury.org/atom.xml"
20}