changing IO Scheduler

[informant]

hi sumsum,

shure, the config of /etc/fstab on host/cluster/node is:

# <file system> <mount point> <type> <options> <dump> <pass>
/dev/pve/root / ext3 errors=remount-ro 0 1
/dev/pve/data /var/lib/vz ext3 defaults 0 1
UUID=5a942cfb-0e70-4ee0-b23a-7bf7a07b0d20 /boot ext3 defaults 0 1
/dev/pve/swap none swap sw 0 0
proc /proc proc defaults 0 0

it´s the default config after install with install cd/iso of proxmox.

regards

 

[informant]

@sumsum, do you have any ideas or a other user?

at the moment we haven´t found a solution for this problem.

regards

 

[sumsum]

Sorry I don't have solution. Its strange anyway, because several people reporting problems in this forum & it is not reproducable yet.

 

[informant]

right, many people have this problem. but no solution is not so good. we hope, that proxmox find a solution. proxmox can test on our servers, but the answer of owner martin is, you must pay a ticket. ok we buy a ticket, proxmox find a solution, and all oter users have a free fix. on a problem with so many users, proxmox must help free. its bug-support. or what you mean?

regards

 

[tom]

right, many people have this problem. but no solution is not so good. we hope, that proxmox find a solution. proxmox can test on our servers, but the answer of owner martin is, you must pay a ticket. ok we buy a ticket, proxmox find a solution, and all oter users have a free fix. on a problem with so many users, proxmox must help free. its bug-support. or what you mean?

regards

Proxmox VE is open source and free to use. If you want support and help from our team, you need to pay.

If you do not pay, we do not support you directly. Its very simple and explained in our wiki:
http://pve.proxmox.com/wiki/Get_support

 

[informant]

hi tom,

sorry for the entrys, we have get a info, that´s a bug of the last openvz kernel. we must wait for a fix of openvz kernel to solve this problem, sorry for "spam" and regards.

 

[tom]

hi tom,

sorry for the entrys, we have get a info, that´s a bug of the last openvz kernel.

that is just a guess so far - lets wait till someone tested with latest adaptec firmware/drivers.

 

[Raymond Burns]

I thought this error was confirmed in relation to having the ext4 filesystem.

 

[tom]

I thought this error was confirmed in relation to having the ext4 filesystem.

there is still no test-case to reproduce this issue.

 

[gkovacs]

there is still no test-case to reproduce this issue.

While there is no reproducible test-case, we do have a single specific VE that - when backed up via vzdump - reliably reproduces the issue on any of our cluster nodes.
At the moment we run this VE under Hyper-V to prevent it from bringing down our nodes every night (under Hyper-V the problem does not appear).

You are welcome to log in and check it out if you choose, feel free to PM/email me.

 

[tom]

did you test updated adaptec driver and firmware already?

 

[gkovacs]

did you test updated adaptec driver and firmware already?

We already have the latest supported driver for our controller in the kernel:

modinfo aacraid
filename:       /lib/modules/2.6.32-19-pve/kernel/drivers/scsi/aacraid/aacraid.ko
version:        1.2-1[29900]custom

As for firmware, one of our controllers run 18512, the other 19076 (latest is 19109).
Both nodes exhibited the issue several times.

 

[tom]

google tells that using aacraid 1.2.1-30200 solved similar issues on other rhel based distros -try it.

 

[tom]

see also http://forum.proxmox.com/threads/94...ter-reset-request-SCSI-hang?p=73476#post73476

 

[informant]

hi tom,

thanks for this nice information. we do update one of our servers today and test this driver.
can you this driver impelment in next test version of proxmox, well in last it´s the old version implemented (1.2.1-29900).?

regards

 

[informant]

hi all,

we have debian wheezy and we download debian 7 driver from https://www.adaptec.com/adapteccom/templates/license-agree.aspx
we get the error message:

dkms add -m aacraid -v 1.2.1.30200

Creating symlink /var/lib/dkms/aacraid/1.2.1.30200/source ->
                 /usr/src/aacraid-1.2.1.30200

DKMS: add completed.
root@titan:~# dkms build -m aacraid -v 1.2.1.30200
Error! echo
Your kernel headers for kernel 2.6.32-20-pve cannot be found at
/lib/modules/2.6.32-20-pve/build or /lib/modules/2.6.32-20-pve/source.
root@titan:~# dkms install -m aacraid -v 1.2.1.30200
Error! echo
Your kernel headers for kernel 2.6.32-20-pve cannot be found at
/lib/modules/2.6.32-20-pve/build or /lib/modules/2.6.32-20-pve/source.
root@titan:~# dkms status
aacraid, 1.2.1.30200: added

what can we do here? please help. very thanks.

regards

google tells that using aacraid 1.2.1-30200 solved similar issues on other rhel based distros -try it.

 

[ebiss]

try:
apt-get install linux-headers-$(uname -r)

 

[informant]

hi,

apt-get install linux-headers-$(uname -r)
Paketlisten werden gelesen... Fertig
Abhängigkeitsbaum wird aufgebaut.
Statusinformationen werden eingelesen.... Fertig
E: Paket linux-headers-2.6.32-20-pve kann nicht gefunden werden.
E: Mittels regulärem Ausdruck »linux-headers-2.6.32-20-pve« konnte kein Paket gefunden werden.

i think, it´s only a listing problem, well proxmox have a own kernel.

regards

 

[ebiss]

Sorry, try:

apt-get install pve-headers-$(uname -r)

 

[informant]

hi,

command works, but after restart the aacraid 1.2.1-30200 driver was cleaned and the old version of proxmox was back.

Sorry, try:

apt-get install pve-headers-$(uname -r)