Search results

node1 May 18 09:28:13 nmz-cl-1 systemd[1]: Stopping pve-ha-lrm.service - PVE Local HA Resource Manager Daemon... May 18 09:28:14 nmz-cl-1 pve-ha-lrm[754284]: received signal TERM May 18 09:28:14 nmz-cl-1 pve-ha-lrm[754284]: got shutdown request with shutdown policy 'migrate' May 18 09:28:14...

Hello, I updated one node and wanted to reboot. In this node was running single VM at the time. HA started migration process but failed with: task started by HA resource agent 2026-05-18 09:29:24 conntrack state migration not supported or disabled, active connections might get dropped...

Hi @nmateo systemd-boot is not bad. You can use ZFS with all options, GRUB cannot, but GRUB have option to lock with password. So use whatever you want. As of all age laws you must look forward - better to have one working solution than multi-solution. As for programs developers you don't need...

I`m tracking Ceph mailing list and from it I assume developers leaves testing stage to customers. Safe upgrade should be done to X.Y.2+ version only.

Try to turn off write cache.

I recommend SoftEther VPN

If I wanted to trick VPS provider I would look does he use ARP. In that case I would use L2 VPN (wireguard is only L3): Cloudzy VPS eth0 -> bridge with VPN -> VPN L2 -> Proxmox -> bridge with VPS -> VPS

Did it worked for short ?

VM OS ignored shutdown or you haven't installed qemu agent. If you really need to stop VM you can use "qm stop" to kill it.

ZFS is not Ceph. You can use up to 100%

VNet. Personally I`m not sure but lets try.

Try to enable "VLAN Aware"

Can you show VM network configuration ?

5 MON - A.I suggest to use this number only on really huge setup. 1 MON - on single ceph machine. 2 MON - no quorum 3 MON - regular setup 4 MON - no quorum? Try to lower MON setup to 3.

I suggest you take a look at these links https://www.osris.org/article/2019/03/01/ceph-osd-site-affinity https://ceph.io/en/news/blog/2015/crushmap-example-of-a-hierarchical-cluster-map/

How OSD lose PG objects ? Who cause it ?

In maintenance time I set noout norebalance norecover flags before OSD/server shutdown. It stops from moving data around others OSD. In some Ceph talks was mentioned that single HDD can impact all cluster event HDD SMART will not show any evidence of coming HDD death. So you must track of disk...

Proxmox installation is not so rich with option. Try to install debian as you wish and then convert to Proxmox.

Hi, have you tried to run 'zpool import -d /dev/disk/by-id' to see what zfs sees ?

You have to ask what Ceph will solve for you. Scalable, failure .... If you decide to use Ceph in my short experience and other people suggestions: 1. Get fast network as possible. Network latency have big role. 2. Get SSD/NVME enterprise grade to survive the load. 3. More CPU cores not always...