Search results

Sure you will not like to have a 50% quorum. That will give you split brain problems. Split brain will lead to VM's running twice and destroying their images, so don't do it. Don't ever think about a 50% quorum. you can setup for example a fifth node on a small box just as an quorum node. Even...

Is the switch port set to autonegotiating?

your pool is full : rpool 1.63T 398G 73K /rpool rpool/KVM 1.62T 398G 57.5K /rpool/KVM one hint: never fill up a COW Filesystem like ZFS, 100% full is really bad, even no deletion is then possible, for good performance better keep it under 70-80 %

did you edit or change in other way the .ssh/authorized_keys file on that node? .ssh/authorized_keys should be a symbolic link: root@gar-ha-kvm11:~# ll .ssh/authorized_keys lrwxrwxrwx 1 root root 29 Mar 29 15:12 .ssh/authorized_keys -> /etc/pve/priv/authorized_keys root@gar-ha-kvm11:~#

Hi, in your setup only vmbr0 and vmbr3 did have an bridge port. Also with OVS setup you do not need to create more than one bridge, Also: Don't mix linuxbridge setup with ovs

with ZVOL (and CEPH also) Format is of course locked to RAW. It is an image which is created with a ZVOL so only RAW is possible. It is thin provisioned by ZFS means, so QCOW is not necessary anyway. RAM: you mean the zfs_arc_max parameter? The size should work but more memory for ARC is always...

Not as I know, just the way I described a virtual IP which is failing over on the NFS Heads

I think you should add it as "ZFS", then you have ZVOL Images

Not necessarily NFS V3 as well as V4.x should work

that's very unusual, do you know the vendor of the NFS share server? Usually for HA is the following: 2 head with own IP 1 virtual IP which fails over between the Heads So you usually use the virtual IP

is that really two NFS Shares? or two IP's for same share?

what do you mean with two IP's for nfs share? Is it a NFS Storage with high availability? Usually there is one virtual IP for the HA Service

First off: As long as you don't have RA's sent from router you will only have link local IPV6 adresses. They look like: inet6-Adresse: fe80::......... A real IPV6 adress will start with 2xxx::....... A security auditor should know that, or he/she is not worth his money. if you have to...

There is no need to disable IPV6 as long as you have no sluggish ipv6 connectivity. Just let the system jobs do its work.

NUT - Network UPS Tools will be an good alternative http://networkupstools.org/

sounds lilke the empty line between the post-up echo and the first post-up iptables is the problem, the iptables commands will probably never be called.

Look into the documentation for zpool command. Something like : zpool create tank mirror dev1 dev2 mirrot dev3 dev4

I don't have such a large number of containers. Maybe you can script it? proxmox has a nice API

I don't know what your problem is, I don bulk migrates for maintenance, the VM's do live, the LXC containers do stop migration as they should. I never configured something specially regarding this.

LXC will be migrated in stop mode of course (no live migration for lxc)