3 navsection: installguide
7 Copyright (C) The Arvados Authors. All rights reserved.
9 SPDX-License-Identifier: CC-BY-SA-3.0
12 {% include 'notebox_begin_warning' %}
13 @crunch-dispatch-slurm@ is only relevant for on premises clusters that will spool jobs to Slurm. Skip this section if you use LSF or if you are installing a cloud cluster.
14 {% include 'notebox_end' %}
16 Containers can be dispatched to a Slurm cluster. The dispatcher sends work to the cluster using Slurm's @sbatch@ command, so it works in a variety of Slurm configurations.
18 In order to run containers, you must run the dispatcher as a user that has permission to set up FUSE mounts and run Docker containers on each compute node. This install guide refers to this user as the @crunch@ user. We recommend you create this user on each compute node with the same UID and GID, and add it to the @fuse@ and @docker@ system groups to grant it the necessary permissions. However, you can run the dispatcher under any account with sufficient permissions across the cluster.
20 We will assume that you have Slurm and munge running.
22 h3. Sample Slurm configuration file
24 Here's an example @slurm.conf@ for use with Arvados:
28 ControlMachine=<span class="userinput">ClusterID.example.com</class>
32 StateSaveLocation=/tmp
33 SlurmdSpoolDir=/tmp/slurmd
34 SwitchType=switch/none
36 SlurmctldPidFile=/var/run/slurmctld.pid
37 SlurmdPidFile=/var/run/slurmd.pid
38 ProctrackType=proctrack/pgid
41 TaskPlugin=task/affinity
52 SchedulerType=sched/backfill
54 SelectType=select/linear
62 JobCompType=jobcomp/none
64 JobAcctGatherType=jobacct_gather/none
68 PartitionName=DEFAULT MaxTime=INFINITE State=UP
70 NodeName=compute[0-255]
71 PartitionName=compute Nodes=compute[0-255] Default=YES Shared=YES
75 h3. Slurm configuration essentials
77 Whenever you change this file, you will need to update the copy _on every compute node_ as well as the controller node, and then run @sudo scontrol reconfigure@.
79 *@ControlMachine@* should be a DNS name that resolves to the Slurm controller (dispatch/API server). This must resolve correctly on all Slurm worker nodes as well as the controller itself. In general Slurm is very sensitive about all of the nodes being able to communicate with the controller _and one another_, all using the same DNS names.
81 *@SelectType=select/linear@* is needed on cloud-based installations that update node sizes dynamically, but it can only schedule one container at a time on each node. On a static or homogeneous cluster, use @SelectType=select/cons_res@ with @SelectTypeParameters=CR_CPU_Memory@ instead to enable node sharing.
83 *@NodeName=compute[0-255]@* establishes that the hostnames of the worker nodes will be compute0, compute1, etc. through compute255.
84 * There are several ways to compress sequences of names, like @compute[0-9,80,100-110]@. See the "hostlist" discussion in the @slurm.conf(5)@ and @scontrol(1)@ man pages for more information.
85 * It is not necessary for all of the nodes listed here to be alive in order for Slurm to work, although you should make sure the DNS entries exist. It is easiest to define lots of hostnames up front, assigning them to real nodes and updating your DNS records as the nodes appear. This minimizes the frequency of @slurm.conf@ updates and use of @scontrol reconfigure@.
87 Each hostname in @slurm.conf@ must also resolve correctly on all Slurm worker nodes as well as the controller itself. Furthermore, the hostnames used in the configuration file must match the hostnames reported by @hostname@ or @hostname -s@ on the nodes themselves. This applies to the ControlMachine as well as the worker nodes.
90 * In @slurm.conf@ on control and worker nodes: @ControlMachine=ClusterID.example.com@
91 * In @slurm.conf@ on control and worker nodes: @NodeName=compute[0-255]@
92 * In @/etc/resolv.conf@ on control and worker nodes: @search ClusterID.example.com@
93 * On the control node: @hostname@ reports @ClusterID.example.com@
94 * On worker node 123: @hostname@ reports @compute123.ClusterID.example.com@
96 h3. Automatic hostname assignment
98 The API server will choose an unused hostname from the set given in @application.yml@, which defaults to @compute[0-255]@.
100 If it is not feasible to give your compute nodes hostnames like compute0, compute1, etc., you can accommodate other naming schemes with a bit of extra configuration.
102 If you want Arvados to assign names to your nodes with a different consecutive numeric series like @{worker1-0000, worker1-0001, worker1-0002}@, add an entry to @application.yml@; see @/var/www/arvados-api/current/config/application.default.yml@ for details. Example:
103 * In @application.yml@: <code>assign_node_hostname: worker1-%<slot_number>04d</code>
104 * In @slurm.conf@: <code>NodeName=worker1-[0000-0255]</code>
106 If your worker hostnames are already assigned by other means, and the full set of names is known in advance, have your worker node bootstrapping script send its current hostname, rather than expect Arvados to assign one.
107 * In @application.yml@: <code>assign_node_hostname: false</code>
108 * In @slurm.conf@: <code>NodeName=alice,bob,clay,darlene</code>
110 If your worker hostnames are already assigned by other means, but the full set of names is _not_ known in advance, you can use the @slurm.conf@ and @application.yml@ settings in the previous example, but you must also update @slurm.conf@ (both on the controller and on all worker nodes) and run @sudo scontrol reconfigure@ whenever a new node comes online.