...
+The dispatcher normally runs on the same host/VM as the API server.
+h2. Perl SDK dependencies
-The dispatcher normally runs on the same host/VM as the API server.
+Install the Perl SDK on the controller.
-h4. Perl SDK dependencies
+* See "Perl SDK":{{site.baseurl}}/sdk/perl/index.html page for details.
-* @apt-get install libjson-perl libwww-perl libio-socket-ssl-perl libipc-system-simple-perl@
+h2. Python SDK dependencies
-Add this to @/etc/apt/sources.list@
+Install the Python SDK and CLI tools on controller and all compute nodes.
-@deb http://git.oxf.freelogy.org/apt wheezy main contrib@
+* See "Python SDK":{{site.baseurl}}/sdk/python/sdk-python.html page for details.
-Then
+h2(#slurm). Set up SLURM
-@apt-get install libwarehouse-perl@
+On the API server, install SLURM and munge, and generate a munge key.
-h4. Python SDK dependencies
+On Debian-based systems:
-On controller and all compute nodes:
+<notextile>
+<pre><code>~$ <span class="userinput">sudo /usr/bin/apt-get install slurm-llnl munge</span>
+~$ <span class="userinput">sudo /usr/sbin/create-munge-key</span>
+</code></pre>
+</notextile>
-* @apt-get install python-pip@
-* @pip install --upgrade virtualenv arvados-python-client@
+On Red Hat-based systems, "install SLURM and munge from source following their installation guide":https://computing.llnl.gov/linux/slurm/quickstart_admin.html.
-h4. Likely crunch job dependencies
+Now we need to give SLURM a configuration file in @/etc/slurm-llnl/slurm.conf@. Here's an example:
-On compute nodes:
+<notextile>
+<pre>
+ControlMachine=uuid_prefix.your.domain
+SlurmctldPort=6817
+SlurmdPort=6818
+AuthType=auth/munge
+StateSaveLocation=/tmp
+SlurmdSpoolDir=/tmp/slurmd
+SwitchType=switch/none
+MpiDefault=none
+SlurmctldPidFile=/var/run/slurmctld.pid
+SlurmdPidFile=/var/run/slurmd.pid
+ProctrackType=proctrack/pgid
+CacheGroups=0
+ReturnToService=2
+TaskPlugin=task/affinity
+#
+# TIMERS
+SlurmctldTimeout=300
+SlurmdTimeout=300
+InactiveLimit=0
+MinJobAge=300
+KillWait=30
+Waittime=0
+#
+# SCHEDULING
+SchedulerType=sched/backfill
+SchedulerPort=7321
+SelectType=select/cons_res
+SelectTypeParameters=CR_CPU_Memory
+FastSchedule=1
+#
+# LOGGING
+SlurmctldDebug=3
+#SlurmctldLogFile=
+SlurmdDebug=3
+#SlurmdLogFile=
+JobCompType=jobcomp/none
+#JobCompLoc=
+JobAcctGatherType=jobacct_gather/none
+#
+# COMPUTE NODES
+NodeName=DEFAULT
+PartitionName=DEFAULT MaxTime=INFINITE State=UP
+PartitionName=compute Default=YES Shared=yes
+
+NodeName=compute[0-255]
+PartitionName=compute Nodes=compute[0-255]
+</pre>
+</notextile>
-* @pip install --upgrade pyvcf@
+h3. SLURM configuration essentials
-h4. Redis
+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@.
-On controller:
+*@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.
-* @apt-get install redis-server@
+*@NodeName=compute[0-255]@* establishes that the hostnames of the worker nodes will be compute0, compute1, etc. through compute255.
+* 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.
+* 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@.
-h4. Crunch user account
+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.
-On compute nodes and controller:
+For example:
+* In @/etc/slurm-llnl/slurm.conf@ on control and worker nodes: @ControlMachine=uuid_prefix.your.domain@
+* In @/etc/slurm-llnl/slurm.conf@ on control and worker nodes: @NodeName=compute[0-255]@
+* In @/etc/resolv.conf@ on control and worker nodes: @search uuid_prefix.your.domain@
+* On the control node: @hostname@ reports @uuid_prefix.your.domain@
+* On worker node 123: @hostname@ reports @compute123.uuid_prefix.your.domain@
-* @adduser crunch@
+h3. Automatic hostname assignment
-The crunch user should have the same UID, GID, and home directory on all compute nodes and on the controller.
+If your worker node bootstrapping script (see "Installing a compute node":install-compute-node.html) does not send the worker's current hostname, the API server will choose an unused hostname from the set given in @application.yml@, which defaults to @compute[0-255]@.
-h4. Repositories
+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.
-Crunch scripts must be in Git repositories in @/var/cache/git/*/.git@ (or whatever is configured in @services/api/config/environments/production.rb@).
+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:
+* In @application.yml@: <code>assign_node_hostname: worker1-%<slot_number>04d</code>
+* In @slurm.conf@: <code>NodeName=worker1-[0000-0255]</code>
-h4. Importing commits
+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 (see "Installing a compute node":install-compute-node.html) send its current hostname, rather than expect Arvados to assign one.
+* In @application.yml@: <code>assign_node_hostname: false</code>
+* In @slurm.conf@: <code>NodeName=alice,bob,clay,darlene</code>
-@services/api/script/import_commits.rb production@ must run periodically. Example @/var/service/arvados_import_commits/run@ script for daemontools or runit:
+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.
-<pre>
-#!/bin/sh
-set -e
-while sleep 60
-do
- cd /path/to/arvados/services/api
- setuidgid www-data env RAILS_ENV=production /usr/local/rvm/bin/rvm-exec 2.0.0 bundle exec ./script/import_commits.rb 2>&1
-done
-</pre>
+h2. Enable SLURM job dispatch
-Once you have imported some commits, you should be able to create a new job:
+In your API server's @application.yml@ configuration file, add the line @crunch_job_wrapper: :slurm_immediate@ under the appropriate section. (The second colon is not a typo. It denotes a Ruby symbol.)
+
+h2. Crunch user account
+
+Run @sudo adduser crunch@. The crunch user should have the same UID, GID, and home directory on all compute nodes and on the dispatcher (API server).
+
+h2. Git Repositories
+
+Crunch scripts must be in Git repositories in the directory configured as @git_repositories_dir@/*.git (see the "API server installation":install-api-server.html#git_repositories_dir).
+
+Once you have a repository with commits -- and you have read access to the repository -- you should be able to create a new job:
<pre>
read -rd $'\000' newjob <<EOF; arv job create --job "$newjob"
{"script_parameters":{"input":"f815ec01d5d2f11cb12874ab2ed50daa"},
"script_version":"master",
- "script":"hash"}
+ "script":"hash",
+ "repository":"arvados"}
EOF
</pre>
ArgumentError: Specified script_version does not resolve to a commit
</pre>
-h4. Running jobs
+h2. Running jobs
* @services/api/script/crunch-dispatch.rb@ must be running.
* @crunch-dispatch.rb@ needs @services/crunch/crunch-job@ in its @PATH@.
-* @crunch-job@ needs @sdk/perl/lib@ and @warehouse-apps/libwarehouse-perl/lib@ in its @PERLLIB@
-* @crunch-job@ needs @ARVADOS_API_HOST@ (and, if necessary in a development environment, @ARVADOS_API_HOST_INSECURE@)
+* @crunch-job@ needs the installation path of the Perl SDK in its @PERLLIB@.
+* @crunch-job@ needs the @ARVADOS_API_HOST@ (and, if necessary in a development environment, @ARVADOS_API_HOST_INSECURE@) environment variable set.
Example @/var/service/arvados_crunch_dispatch/run@ script:
<pre>
#!/bin/sh
set -e
+
+rvmexec=""
+## uncomment this line if you use rvm:
+#rvmexec="/usr/local/rvm/bin/rvm-exec 2.1.1"
+
export PATH="$PATH":/path/to/arvados/services/crunch
-export PERLLIB=/path/to/arvados/sdk/perl/lib:/path/to/warehouse-apps/libwarehouse-perl/lib
export ARVADOS_API_HOST={{ site.arvados_api_host }}
export CRUNCH_DISPATCH_LOCKFILE=/var/lock/crunch-dispatch
+# This is the path to docker on your compute nodes. You might need to
+# change it to "docker", "/opt/bin/docker", etc.
+export CRUNCH_JOB_DOCKER_BIN=docker.io
+
fuser -TERM -k $CRUNCH_DISPATCH_LOCKFILE || true
## Only if your SSL cert is unverifiable:
cd /path/to/arvados/services/api
export RAILS_ENV=production
-exec /usr/local/rvm/bin/rvm-exec 2.0.0 bundle exec ./script/crunch-dispatch.rb 2>&1
+exec $rvmexec bundle exec ./script/crunch-dispatch.rb 2>&1
</pre>
projects / arvados.git / blobdiff