2024-06-26T07:05:59.347Z

<Innocent T.> Thanks a lot.

2024-06-26T15:20:40.229Z

<Raghu> hello, Good morning
Reference : <https://ceph-storage.slack.com/archives/C04SNUBD2M6/p1718130486614289>
we are currently using the following command to bootstrap the cluster
```/usr/sbin/cephadm  --no-cgroups-split --image=...... bootstrap --mon-ip=10.11.22.33  --allow-fqdn-hostname  --cluster-network=10.11.0.0/16 --allow-overwrite --log-to-file --skip-dashboard --skip-monitoring-stack```
During this process , service files are created and the unit.run file does not have --cgroup=split in it. This is good.
when i add the new machines in to the cluster and deploy more MONs, MGRs, RGWs and crash service, all of them still have --cgroup=split in the unit.run service files. we have to manually go there and cleanup the config, shutdown the container instances and restart the ceph services on them.
Is there anything that we can do so that we dont have to do this manual cleanup for any new daemons deployed on the cluster ?
thank you

2024-06-26T18:29:04.170Z

<Adam King> There is a separate `mgr/cephadm/cgroups_split` option that can be set to false after the cluster has been bootstrapped that should make daemons cephadm deploys not include the `--cgroups-split` in it. I had thought providing the  `--no-cgroups-split`bootstrap flag would make it set that option to false but looking at the code it appears not to be doing so.

2024-06-26T19:19:14.075Z

<Raghu> thank you, this seems like a bug as well, i will create a tracker for this .