Galaxy Maintenance Docker Container

A Docker image containing the Galaxy packages and tools needed to parse the galaxy.yml config file and talk to the Postgresql database, including psql, gxadmin and the Galaxy maintenance scripts cleanup_datasets.py, pgcleanup.py, and a modified version of the maintenanc.sh script.

Assumptions

This documentation assumes Galaxy was installed using galaxy as the release name and galaxy as the namespace. If you used different values, you will need to adjust the commands accordingly.

Usage

The easiest way to use the container is to run it as a cron job with a scheduled time that will never occur (Feb 30th in this example). Then trigger the job manually when needed. This example runs a simple Bash script that waits for a watch file to be created, then exits when it is created.

helm upgrade galaxy -n galaxy galaxy/galaxy --reuse-values -f templates/admin.yaml
kubectl create job -n galaxy --from=cronjob/galaxy-cron-admin admin-manual

The pod that is created with have a random string appeneded to admin-manual that we need to find to connect to the pod.

pod=$(kubectl get pods -n galaxy -l job-name=admin-manual -o jsonpath='{.items[0].metadata.name}')
kubectl exec -n galaxy -it $pod -- /bin/bash

To terminate the job, create the watch file.

kubectl exec -n galaxy $pod -- touch /galaxy/server/database/tmp/debug_watch.lock
# Wait for the job to terminate
kubectl delete job admin-manual

Example CronJob

cronJobs:
  admin:
    enabled: false
    schedule: "0 0 30 2 *"
    image:
      repository: ksuderman/galaxy-maintenance
      tag: "0.11.5"
    extraEnv:
      - name: WATCH
        value: /galaxy/server/database/tmp/debug_watch.lock
      - name: LOG
        value: /galaxy/server/database/debug.log
      - name: PGHOST
        value: |-
          {{ include "galaxy-postgresql.fullname" . }}
      - name: PGUSER
        value: |-
          {{ .Values.postgresql.galaxyDatabaseUser }}
      - name: PGDATABASE
        value: galaxy
      - name: GALAXY_CONFIG_FILE
        value: /galaxy/server/config/galaxy.yml
      - name: GALAXY_ROOT
        value: /galaxy/server
      - name: GALAXY_LOG_DIR
        value: |-
          {{ .Values.persistence.mountPath}}/tmp
    command:
      - /usr/local/bin/main.sh
    extraFileMappings:
      /usr/local/bin/main.sh:
        mode: "0755"
        content: |
          #!/usr/bin/env bash
          # The .pgpass file needs to be owned by the Galaxy user with 0600
          # permissions.  However, ConfigMap files are always owned by root.  So we
          # need to copy the file to the Galaxy user's home directory and
          # change the ownership and permissions.
          cp /tmp/pgpass /home/galaxy/.pgpass
          sudo chown galaxy:galaxy /home/galaxy/.pgpass
          chmod 0600 /home/galaxy/.pgpass
          while [[ ! -e $WATCH ]] ; do
            # Sleep until the $WATCH file is created
            date | tee $LOG
            sleep 15
          done
          echo "Kill file $WATCH detected." | tee $LOG
          rm $WATCH
      /tmp/pgpass:
        tpl: true
        mode: "0644"
        content: |-
          {{- (include "galaxy-postgresql.fullname" .) }}:5432:*:{{- .Values.postgresql.galaxyDatabaseUser }}:{{- (include "galaxy.galaxyDbPassword" .) }}
      /galaxy/server/config/galaxy.yml:
        tpl: true
        content: |-
          {{- index .Values.configs "galaxy.yml" | toYaml | nindent 4 }}