4 node cluster. Bonded (LACP) 10Gb NIC backend. Using Ceph storage (6 x nvme 2TB, 4 OSDs per nvme). All other operations seem quite fast. VM disk is stored on ceph.
The state transfer is going soooooo slooooow. Not sure if this is expected or something I've mucked up, but seeing all the warnings around the Perl module for migration makes me concerned. Running latest updates for 8.4.1 as of this morning. Just ramping up to put this cluster into production but I need to resolve this super slow speed. I can't imagine HA would even work properly with this speed!
The state transfer is going soooooo slooooow. Not sure if this is expected or something I've mucked up, but seeing all the warnings around the Perl module for migration makes me concerned. Running latest updates for 8.4.1 as of this morning. Just ramping up to put this cluster into production but I need to resolve this super slow speed. I can't imagine HA would even work properly with this speed!
Code:
2025-05-22 11:27:39 migration active, transferred 27.7 MiB of 4.0 GiB VM-state, 244.0 KiB/s
Use of uninitialized value in string ne at /usr/share/perl5/PVE/QemuMigrate.pm line 1324.
Use of uninitialized value $last_vfio_transferred in string ne at /usr/share/perl5/PVE/QemuMigrate.pm line 1324.