Shoveler installation and monitoring
Shoveler is simple to install and is currently running a test version on cms-rucio-services.gridpp.rl.ac.uk as set up by Katy. Now it is time to bring into production where it will be installed and managed as a full production service.
When testing first started, I was asked to have a VM created to run Shoveler eventually in production…this was 2 years ago. In case this machine still exists, here is the name:
shoveler-monit01
This was created by Rob Harper in May 2022. It has an inventory personality.
For more information about what Shoveler does and how to install it, please see the documentation here: GitHub - opensciencegrid/xrootd-monitoring-shoveler: XRootD monitoring flow packet shoveler
At time of writing the advice is to install the latest version which is 1.4.0.
To configure Shoveler, please see the config file on the cms-rucio-services machine in this location:
/etc/xrootd-monitoring-shoveler/config.yaml
Also that config is here: https://stfc.atlassian.net/wiki/spaces/X/pages/476577802
To start/stop/status Shoveler, use commands such as:
systemctl start xrootd-monitoring-shoveler.service
One issue I have had with Shoveler is that sometimes it stops sending data to the Collector, but at the same time the ‘status’ looks healthy. When this happens Shoveler stores records of XRootD in:
/tmp/shoveler-queue
But it requires a stop/start to force it to start sending the data again.