It is important to understand the behaviour of the ceph services and to make the right hardware dimensioning. You can also limit the resource consumption of each service to prevent the memory consumption of the ceph osd from affecting the ceph mon service on the same node. The resource required for the ceph mds depends heavily on the expected workload. We use the setup you describe very often. Hope this helps you one step further, Joachim Joachim Kraftmayer CEO +49 89 2152527-21 <https://call.ctrlq.org/+49%2089%202152527-21> joachim.kraftmayer@xxxxxxxxx www.clyso.com Hohenzollernstr. 27, 80801 Munich Utting a. A. | HR: Augsburg | HRB: 25866 | USt. ID-Nr.: DE2754306 Am Mo., 26. Aug. 2024 um 09:48 Uhr schrieb Janne Johansson < icepic.dz@xxxxxxxxx>: > Den fre 23 aug. 2024 kl 18:30 skrev Phong Tran Thanh < > tranphong079@xxxxxxxxx>: > > > > This is my first time setting up a Ceph cluster for OpenStack. Will > > running both the Mon service and the OSD service on the same node affect > > maintenance or upgrades in the future? While running Mon, MDS, and OSD > > services on the same node does offer some benefits, could someone provide > > me with additional advice? > > OSDs can use HUGE amounts of ram when doing internal consistency > checks and things like that, so co locating the mon(s) with the OSDs > would mean the host can end up swapping or even dying from lack of > RAM, and while the machine is heading this way the mon part will > probably not run well either, so the OSDs could starve the mon and you > would be having two problems at the same time if this ever occurs. > > -- > May the most significant bit of your life be positive. > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx