Description of problem: sap_*_preconfigure roles install @server group and that brings in lots of additional services on minimal installations. It might be a good idea to disable some of the unneeded services even if many of them are no-op. For instance, iscsi related services should be no-op but to make it explicit that no iscsi is in use those services could be disabled. Perhaps slightly more importantly, disabling mlocate-updatedb.timer would prevent scanning all files and mounts once per day and thus potentially causing slight IO load/latency, although the unit should run on very low prio. These are mere suggestion for potential tiny improvement, please feel free to accept or reject and close this BZ if not deemed worth the effort. Thanks.
FWIW, these might be candidates to disable but as said most of them should be no-op: - iscsid.socket - iscsi-onboot.service - iscsi.service - iscsiuio.socket - mdmonitor.service - mlocate-updatedb.timer - nfs-client.target - rpcbind.service - rpcbind.socket - sssd-kcm.socket - sssd.service - vdo.service
I am not sure if disabling such services can have negative consequences for SAP. Bernd has already mentioned in 2078423 that we are trying to bring the roles as far as possible in line with the applicable SAP notes. @Bernd, what is your opinion on this?
Disabling services is probably something we could mention in https://access.redhat.com/articles/6452131 after verifying that these services can indeed be disabled.
(In reply to Bernd Finger from comment #3) > Disabling services is probably something we could mention in > https://access.redhat.com/articles/6452131 after verifying that these > services can indeed be disabled. I suppose that it will take while to verify that these services can really be disabled. Bernd, Is it worth fixing it? if so can we fix this problem in rhel-8.8 update? Thank you!
Perhaps one additional thing worth mentioning is that if we would be one day able to install only @minimal-environment instead of @server that would largely reduce the number of installed/enabled service by itself. So in case that is something that might happen perhaps this could be checked after that. But if that's not possible then disabling at least mlocate-updatedb.timer might be warranted. But this is just a suggestion, I'll leave it up to you to do decide. Thanks.