[SOLVED] error on ubuntu server 16.04

[abranca]

Hello, on ubuntu server machines installed on proxmox 4.4 I find this kind of errors and I can not figure out what's wrong. can you help me?

ata3.00: status: { DRDY ERR }
ata3.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
ata3.00: failed command: WRITE DMA EXT

Vm is on a virtual hard disk with ext4 partitions.

The physical machine has two hd configured with raid1 during the operating system installation.

 

[marsian]

Hi, which IDE driver did you select?

 

[abranca]

hi marsian, the vm disk is sata

 

[marsian]

We noticed several issues with virtual drives if the driver is not "virtIO", causing errors like the one above on the guest system. Try changing the driver + reboot and see if it's gone

 

[abranca]

can i use this procedure?

Shutdown VM
Add new VirtIO HDD to your server (like a 1gb hdd)
Start VM
Use the virtio drivers for scsi controllers
Shutdown VM
Delete all harddisks
Re-add all harddisks with VIRTIO
Go to options, select your bootable HDD again.

 

[abranca]

I used the above procedure on a test machine that is usually off. I'm doing tests on this machine.
Other machines that are experiencing the error, when restarting no longer recognize the hard disk as bootable, so I must use testdisk to restore the partitions and reinstall grub. Before making a change to the test run on this temporary machine.

 

[udo]

can i use this procedure?

Shutdown VM
Add new VirtIO HDD to your server (like a 1gb hdd)
Start VM
Use the virtio drivers for scsi controllers
Shutdown VM
Delete all harddisks
Re-add all harddisks with VIRTIO
Go to options, select your bootable HDD again.

Hi,
adding first a small disk is for windows only - to install the driver... this is not nessessary for an actual linux!
You should only take a look at /etc/fstab - normaly there are used uuids and all is fine.

Simply remove disk and edit -> add as needed type again.

But your issue sounds not realy that an change from scsi (scsi controller virtio?) to virtio will help.
What kind of underlying storage do you use? For some special types* you should not use the default cache (no cache) but other cache types worked, like write through.
* like LVM with 4K-Sector-size

Udo

 

[abranca]

Hi Udo, thank you for your explanation.
The virtual machine is configured on a virtual hd set as sata, for the cache I used the default (no cache).
On the virtual machine runs ubuntu server 16.04 with two partitions: swap and root formatted in ext4 without using lvm.
Maybe I was wrong about the installation or configuration procedure.
Add pictures with details of the virtual machine configuration.

Regarding the fstab I have to check that the virtual machine is correct?

 

[udo]

Hi Udo, thank you for your explanation.
The virtual machine is configured on a virtual hd set as sata, for the cache I used the default (no cache).
On the virtual machine runs ubuntu server 16.04 with two partitions: swap and root formatted in ext4 without using lvm.
Maybe I was wrong about the installation or configuration procedure.

Hi,
looks quite normal.

Regarding the fstab I have to check that the virtual machine is correct?

yes, I mean the fstab of the VM - if there an enty with "/dev/sda1 / ext4 ..." it will not work with virtio because the name was vda1 then.

The VM was installed on the proxmox host or migrated from another system? (but should work too).

Are there any logentrys on the host, related to the IO?

Udo

 

[abranca]

I attach the fstab of the above VM.
Don't use / dev / sdax but uuid

The VM was installed on the proxmox host.

 

[abranca]

I recovered and re-imported disks with virtio controller and the problem seems to be resolved. Thank you very much!