Proxmox KVM Machine crashing - journal commit I/O error

[deffe]

Hello,

I have changed the entire dedicated server, new cpu, motherboard ram and harddrives are brand spanking new. I thought I had a hardware error but this is something else releated to the KVM itself.

I have updated to the latest proxmox version 3.4-11

The problem is that my KVM machine all of the sudden started to crash, it throws out "journal commit I/O error"
and then the filesystem is locked to read only mode.

I have tested to run it with cache: none/default, writethrough and write back.

Everytime it crashes witch it does often I have to go into fsck on the KVM and fix brokes nodes and similar before reboot again.
I am totally lost, could you please tell me what could possibly be wrong?

The load is not high. I have a max load of 1-2mb/sec on the harddrives and the CPU is running almost at idle (10%).

This started last thursday and it seems to be accelerating. Also it seems like when the mysql starts to load a bit harder that is when it crashes, but as I said the total load on the server is very low and usually below 10%.

Your help would be much appreciated.

 

[spirit]

hi,
what is your server model, and disk controller ?
do you use kerel 2.6.32 or 3.10 ?

what is your vm config . (can post the /etc/pve/qemu-server/vmid.conf ?)

inside the guest, what is the distro, kernel version. Do you have disable barrier on the disk ?

 

[deffe]

hi,
what is your server model, and disk controller ?
do you use kerel 2.6.32 or 3.10 ?

what is your vm config . (can post the /etc/pve/qemu-server/vmid.conf ?)

inside the guest, what is the distro, kernel version. Do you have disable barrier on the disk ?

I am using Hetzner PX90, it is the same as the PX91.
https://www.hetzner.de/se/hosting/produkte_rootserver/px91

Intel® Xeon® E5-1650 v3
Hexa-Core Haswell
incl. Hyper-Threading Technology

• RAM64 GB DDR4 ECC RAM
• Hard Drive2 x 2 TB SATA 6 Gb/s 7200 rpm
  HDD (Software-RAID 1)
  Class Enterprise

The guest is Running CloudLinux 6.5

This is the .conf
cat /etc/pve/qemu-server/102.conf
balloon: 8000
boot: cdn
bootdisk: virtio0
cores: 5
ide2: local:iso/CentOS-6.7-x86_64-LiveCD.iso,media=cdrom,size=697M
keyboard: en-us
machine: pc-i440fx-1.7
memory: 24000
name: whm.nametyper.com
net0: e1000=00:50:56:00:5A:06,bridge=vmbr0
net2: e1000=00:50:56:00:65:47,bridge=vmbr0
onboot: 1
ostype: other
parent: afterCleanup
sockets: 2
virtio0: local:102/vm-102-disk-1.qcow2,format=qcow2,cache=directsync,size=1500G


[afterCleanup]
#Servern har blivit rensat p%C3%A5 skit och massa knasigheter.
balloon: 16000
bootdisk: virtio0
cores: 8
ide2: none,media=cdrom
machine: pc-i440fx-1.7
memory: 32000
name: whm.nametyper.com
net0: e1000=00:50:56:00:5A:06,bridge=vmbr0
net2: e1000=00:50:56:00:65:47,bridge=vmbr0
onboot: 1
ostype: other
snaptime: 1439041779
sockets: 2
virtio0: local:102/vm-102-disk-1.qcow2,format=qcow2,size=1500G