Finished ingestion service as task in rucio task manager (gave to JC to add to argocd, but image is done).
Migrated StoRM. Works, but still need to port data across. Having problems with volume attachments for rse data volume.
Continued drafting ingestion node requirements.
Continuing converting recipe below into helm chart but cloud went down for day.
Made confluence page with requirements for ingestion nodes.
Started looking at running the ingestion service as part of a task in rucio task manager.
Converting recipe below into helm chart.
Got ingestion service working.
StoRM now running on rocky9 machine. Would like to turn this into an image + helm chart with PV so can run it on cluster.
Rough recipe:
sudo yum install git maven java-11-openjdk.x86_64
sudo git clone https://github.com/italiangrid/storm-webdav.git /opt/storm-webdav
sudo chown -R rocky:rocky storm-webdav/
sudo mkdir -p /usr/share/java/storm-webdav/
sudo mv storm-webdav-server.jar /usr/share/java/storm-webdav/
sudo mkdir -p /etc/systemd/system/storm-webdav.service.d/
(copy /usr/lib/systemd/system/storm-webdav.service)
(copy /etc/systemd/system/storm-webdav.service.d)
sudo ln -s /usr/lib/systemd/system/storm-webdav.service storm-webdav.service
sudo systemctl daemon-reload
sudo mkdir -p /etc/storm/webdav
sudo useradd -ms /bin/bash storm
(copy /etc/storm/webdav)
sudo mkdir -p /var/log/storm/webdav/
sudo chown -R storm:storm /var/log/storm/
sudo mkdir -p /var/lib/storm-webdav/work
sudo chown -R storm:storm /var/lib/storm-webdav/
(copy /etc/grid-security/storm-webdav/)
sudo service storm-webdav start
sudo service storm-webdav status
Added persistence to site capabilities API.
Helmified ingestion service + added ability to add persistent volumes to rucio task manager
Migrated Rucio namespace.
Made numerous changes to ingestion service and added docs (https://ska-src-ingestion-ska-telescope-57397dbf79eedf6757468cdbd1a13d7.gitlab.io/srcnet-tools-ingest.html):
Finished migrating monitoring namespace. See https://confluence.skatelescope.org/display/SRCSC/Tech+Debt%3A+Migrate+PVs+to+use+Manila+storage+rather+than+Cinder
Got this to work by deleting PVs and PVCs (retained), but the storage class hasn't migrated. Although it's listed as manila on the server, the provisioner used is still Cinder.
Put monit back online while i work this out. Will need to backup/restore again.
APIs down because they can't see the SKA IAM (doesn't route on the instance but DNS works). Can't log in to Openstack. Can't describe Velero backups (I could yesterday).
Something is awry at RAL.
Mostly just catching up at HQ.
Installed manila. Installed velero.
Can hopefully take backups and restore with different storage class:
https://velero.io/docs/v1.13/restore-reference/#changing-pvpvc-storage-classes
Snapshotted indices yesterday. If this is tested tomorrow this should be OK as volumes are set to retain when unattached.