ATM druid hosts are using custom partman recipes which use /var/lib/druid as mount point as the "data volume". We could switch to standard raid10 recipe, although that would entail moving to /srv (e.g. /srv/druid). Filing a task to explore options, @elukey mentioned new druid hosts need to be put in service soon and this seems like a good time
|operations/puppet||production||+1 -75||install_server: move druid to standard partman recipe|
One option would be to manually create /srv/druid symlinks on the existing installed base and then switch Puppet to use it, with buster reimages and hw refreshes, the underlying remaining uses of /var/lib/druid would vanish over time.
So I checked and the /var/lib/druid dir seems also created by the deb:
elukey@druid1001:~$ dpkg -S /var/lib/druid druid-common, druid-middlemanager, druid-historical: /var/lib/druid
As far as I can see from druid1001 there is no reason why we (as Analytics) couldn't use /srv/druid. Since we are expanding the clusters very soon in T245569, I'd just use Buster for the new nodes and then test the new partition scheme in there. So +1 from my side to go ahead with using /srv/druid!