Mailing List Archive

Dear DRBD/LINSTOR users on PVE,

this is the RC1 of the upcoming 6.0.0 release. There are no dramatic
changes, but by bumping the major version I took the chance to flip
some defaults and remove some very old cruft.

The option 'preferlocal' is now set to 'yes' by default. This means when
you create a VM, the plugin tries to create storage on the node the
'create' command was issued. Previously it could have been that you
selected nodeA in the GUI, created the VM and it ended diskless on nodeA
and with actual storage on nodeB.

Keeping a status cache ('statuscache') for storage pool information is
now enabled by default and the default duration is 1 Minute. This was
available for a long time and users with huge clusters ran into issues.
Unfortunately PVE queries information a lot and the internal PVE cache
that can be used does not solve all issues.

The 'controllervm' (not to confuse with 'controller') setting that was
used for some form of controller HA is gone. We nowadays have ways
better means to implement a HA linstor controller[1].

The 'redundancy' setting code was removed. This was completely dead
code, and was replaced with LINSTOR resource groups and their redundancy
setting a long time ago.

If you have the latest version installed it should be sufficient to
replace the LINSTORPlugin.pm on your system
(/usr/share/perl5/PVE/Storage/Custom) and restart the pvedaemon.service

Please test! If nobody complains this will become 6.0.0 in about a week
and get moved to the repositories.

Regards, rck

GIT:
https://github.com/LINBIT/linstor-proxmox/commit/83b769059539ac661a9fb627b5349f74bedfb789
TGZ:
https://pkg.linbit.com//downloads/connectors/linstor-proxmox-6.0.0-rc.1.tar.gz

[1]
https://linbit.com/drbd-user-guide/linstor-guide-1_0-en/#s-linstor_ha