Search results

Did you copy the key over for proxmox to use for the communication? mkdir -p /etc/pve/priv/ceph cp /etc/ceph/ceph.client.admin.keyring /etc/pve/priv/ceph/ceph-sshd_01.keyring

Pretty sure this is a ceph issue, I've filed a bug with Ceph: http://tracker.ceph.com/issues/10116

I can consistently reproduce a VM lockup in my production environment. I install a CentOS 7 guest on a ceph rbd storage pool, and within that VM, I run: fio --rw=randrw --refill_buffers --norandommap --randrepeat=0 --ioengine=libaio --bs=4k --rwmixread=100 --iodepth=16 --numjobs=16...

Nope, as stated in my original statement, fstrim works within the guest and does as intended. My problem is random deadlocks when I have discard enabled (which facilitates the use of fstrim).

I've been attempting to play around with the discard option in proxmox. When I enable that option, it seems I get random and unrecoverable lockups in the guest VMs (CentOS 7). The lockup doesn't appear to be disk-only, nothing works, console is unresponsive. Usually when I see a disk issue...

Sorry, no, I run a very homogenous environment, Linux only. I just remembered running across that wiki in the past and wasn't sure if the registry hack helped or not, sounds like it doesn't. The only other thing I might suggest is trying the q35 machine type, no clue if that would actually...

Did you use Virtio and the Windows registry hack as documented here: https://pve.proxmox.com/wiki/Paravirtualized_Network_Drivers_for_Windows ?

Your config doesn't look like the https://pve.proxmox.com/wiki/Open_vSwitch wiki. First, you're using OVSPort for the ethernet device, don't think you need to do that. Also, you seem to be missing allow-ovs vmbr1 before your bridge definition. I'd recommend following that wiki.

Ok, looks like you're right, networking came back at boot without it. Perhaps I put it in place but wasn't my actual resolution while I was trying to figure out openvswitch.

Ok, your version looks good, but you didn't mention if you added the "allow-ovs vmbr0"

Are you sure? I just hit that last week when using the latest pve-no-subscription and had to hunt around for the fix. Which is why I included it in the wiki I created for Open vSwitch. I can try removing it from one of my servers in my test lab to verify.

RGManager won't start unless you have any VMs assigned to be HA. What is listed in the <rm> section of /etc/pve/cluster.conf ?

Please verify what version of openvswitch you have installed ... e.g. dpkg -l | grep openvswitch The only other thing I see is the config spirit provided is missing allow-ovs vmbr0 Prior to the vmbr0 definition. The OVS docs state it is required, I've never tried without it.

Did you deploy the Startup Workaround as described on this wiki: https://pve.proxmox.com/wiki/Open_vSwitch#Startup_Workaround ??

The part you said you tried to exclude, you definitely need to exclude. However, that's not the core of your problem, I guess I forgot to mention there is an OVS bug at boot at the moment. I ended up writing a wiki on OVS yesterday since I saw a bunch of people were hitting things I hit. It...

You would need to give an example of how your network is laid out to comment on if your configuration is valid or not. The only time you would assign an ip address to the actual bridge itself with OVS is if you're trying to use an untagged vlan. I've never used the GUI for network...

A working openvswitch with bonding, bridging, and vlans is shown on this thread, along with a work-around for why the interfaces don't come up on boot: http://forum.proxmox.com/threads/19972-Open-vSwitch-and-multicast-issues-%28cluster-keeps-losing-quorum%29

I've done a mac mini proxmox cluster using current-generation minis. My entire howto is here: http://www.jaxlug.net/wiki/2014/07/16 I'm actually not sure how you got as far as you did based on what you've stated due to the EFI hoops you have to jump through, unless of course, you're not using...

You need to use OVSIntPort's to separate out your vlans into interfaces on the host. Do not attempt to mix regular linux bridging, bonding and vlans with OVS, it won't work (you can't assign a vlan to an ovs bond). Also, you could never use a bond directly that was part of a bridge, that...

Any reason not to put the querier on the router instead of the switch? Typically we don't use RVI (Routed Virtual Interface) interfaces on our switches, so you can't do an IGMP querier without an RVI. I enabled it on our Juniper SRX router via set protocols igmp $iface version 2 and it...