Recent content by Mike_TBF

sorry for warming up this old thread, but it is appropriate for my question and probably the place others would look for it: you mentioned using more than one separate cache pools, one per host, in read-only mode. Even the Ceph documentation mentions the possibility of having more than one...

sorry to bother you, mates. I found the cause of the issue. The network adapters are connected to a Zyxel XS1920 switch, and that has a known interoperability issue with Intel X550 based network adapters. It is not enough to upgrade the switch's regular firmware though, but you must install the...

Right now, they are connected to a 10G switch. I did not try a direct connection yet. I have a remote connection to the cluster, but no physical access during the weekend. So I can only test that on Monday.

Hi, this happened to me as well at one point. However, it seems that basically, OSDs do not rely on their device paths. After another reboot of that node they came back, even though the device paths were different. I think I had the journals on the same disks at that time. However, when I look...

Hi everyone, right now I am trying to get Proxmox running on some new Supermicro X10SRM-TF mainboards. They have dual 10G ethernet on board, connected to a Zyxel 10G switch; lspci reports the chipset as Ethernet controller: Intel Corporation Ethernet Controller 10G X550T (rev 01) Networking...

Hi, Proxmox is very "protective" when it comes to virtual disks. Depending on your storage, you cannot simply replace them with something else. On some storage adapters there will be easier / faster ways, but this should work in any case: Create a new VM and add a virtual disk to it that has...

standing by my examples above, I very much believe it would be worse. After all, the VMs would need to got through the 1gbe network when talking to their storage. This would be particularly bad for VMs that deliver storage content to clients, like file servers etc, since all data they read from...

Yes, you need to do this on the CLI. See the Ceph docs here for some pointers on how to choose the right number of PGs and the actual commands to do it: docs.ceph.com/docs/master/rados/operations/placement-groups (unfortunately I am not allowed to post links, so add the "http" yourself. :-) )

No, this is not an error, it is a matter of design philosophy. The Proxmox Wiki assumes your Ceph cluster network to be 10Gbit/s, and thus significantly faster than the Proxmox public network. It also assumes that only your Proxmox hosts will access Ceph. In this case it is probably a good idea...

Hi, I am pretty new to Proxmox with Ceph too, but had a little time to get accustomed to it. Since Proxmox configures Ceph as a "cluster-internal storage" and the vhosts have access to the cluster network, there is no need for the monitors to be reachable from the outside. At least this is...