[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] Re: Failed to create VM using configuration specified by Remus


  • To: zhangninja@xxxxxxxxx, xen-devel@xxxxxxxxxxxxxxxxxxx, keir.fraser@xxxxxxxxxxxxx
  • From: Minjia Zhang <zhangninja@xxxxxxxxx>
  • Date: Fri, 16 Apr 2010 17:09:17 +0800
  • Cc:
  • Delivery-date: Fri, 16 Apr 2010 02:09:59 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; b=sluehZ2lA+XtiERJ9L1agmKPPX9LsYY+bIYlVDBuRPRMvEmmj/y1o0fQG6WC2C4y3e fNitEUAIrLglew5ed3dMms8/zwVYDodIa56HjeXB6cnOzcv6fvHvjhZ1QXFLBdWFwBSM OnBJ7kvIAKhhq7BU+Xy06uOmAzP0nuqJTAZnA=
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>

Hi, Brendan,
Thank you very much for your reply and help. I've tested Remus with xen-unstable-4.0 and the latest linux-2.6.18-xen.hg. ÂThe remus seems to work well. The previous error seems caused by using an old version of linux-2.6.18-xen.hg and I've fixed it.

I encounter some problem when try to run remus with two vms on one singleÂphysicalÂÂmachine,like:
remus vm1 backup
remus vm2 backup
The first command can run correctly, while the second line get error and give the following messages. Can Remus provide two vms (on one physical machine) the fault tolerance simultaneously?

[root@server1 test]# remus 34 server2

ERROR Internal error: Can't create lock file for suspend event channel

WARNING: suspend event channel unavailable, falling back to slow xenstore signalling

Had 0 unexplained entries in p2m table

Â1: sent 64491, skipped 725, delta 3425ms, dom0 75%, target 75%, sent 617Mb/s, dirtied 10Mb/s 1085 pages

Â2: sent 1083, skipped 2, delta 43ms, dom0 100%, target 100%, sent 825Mb/s, dirtied 12Mb/s 16 pages

Â3: sent 15, skipped 1, Start last iteration

PROF: suspending at 1271404630.401318

installing buffer on imq0

RTNETLINK answers: File exists

ERROR Internal error: Suspend request failed

ERROR Internal error: Domain appears not to have suspended

Save exit rc=1

Traceback (most recent call last):

ÂÂFile "/usr/bin/remus", line 359, in ?

ÂÂÂÂrun(cfg)

ÂÂFile "/usr/bin/remus", line 340, in run

ÂÂÂÂfor buf in bufs:

ÂÂFile "/usr/bin/remus", line 277, in postsuspend

ÂÂÂÂbuf.postsuspend()

ÂÂFile "/usr/bin/remus", line 159, in postsuspend

ÂÂÂÂself._setup()

ÂÂFile "/usr/bin/remus", line 185, in _setup

ÂÂÂÂself.rth.talk(req.pack())

ÂÂFile "usr/lib/python2.4/site-packages/xen/remus/netlink.py", line 314, in talk

IOError: error sending message


On Tue, Apr 6, 2010 at 5:30 AM, Brendan Cully <brendan@xxxxxxxxx> wrote:

On Friday, 02 April 2010 at 22:29, Minjia Zhang wrote:
> I fail to use remus with xen-unstable (4.1)+ Linux-2.6.18.8:
>
> [root@server1 img]# xm create rhel5.hvm
> Using config file "./rhel5.hvm".
> Error: Device 768 (tap) could not be connected.
> server1:9000|aio:/dev/vgxen/hvmsnap does not exist
>
> Am I missing something? Any hints will be appreciated.
>
> According to Brendan's instruction on http://dsg.cs.ubc.ca/remus/doc.html
> the config should like following:
> disk = [ 'tap:remus:bkup:9000|aio:/dev/vmdisk/myguest,xvda1,w' ]
>
> According to my understandingïremus doesn't need the support of DRBD, cause
> it can continuously send memory state to backup server as well as write
> operation and network
> packages. So, what I do is to only ensure that /dev/vgxen/hvmsnap existing
> on both the primary and the backup and contain identical data.

That's correct.

> My config file:
> Â 1 import os, re
> Â 2 arch = os.uname()[4]
> Â 3 if re.search('64', arch):
> Â 4 Â Â arch_libdir = 'lib64'
> Â 5 else:
> Â 6 Â Â arch_libdir = 'lib'
> Â 7 kernel = '/usr/lib/xen/boot/hvmloader'
> Â 8 builder = 'hvm'
> Â 9 memory = '512'
> Â14 disk = [ 'tap:remus:server1:9000|aio:/dev/vgxen/hvmsnap,hda,w']
> Â16 vif = ['type=ioemu, mac=00:1c:3e:17:22:13']
> Â17 #boot = 'dc'
> Â18 boot = 'c'
> Â19 name = 'Rhel5-hvm1'
> Â20 acpi = 1
> Â21 apic = 1
> Â22 device_model = '/usr/' + arch_libdir + '/xen/bin/qemu-dm'
> Â23 vnc=1
> Â24 vncdisplay=2
> Â25 sdl=0
> Â26 opengl=1
> Â27 vnclisten="0.0.0.0"
> Â28 vncpasswd=''
> Â29 serial='pty'
> "hvm.hvm" 29L, 975C

This config looks ok to me. You'll get an error like that with pygrub,
which doesn't work with the Remus disk replication yet, but your HVM
config looks correct. I've just tested 32-bit HVM ubuntu on
xen-4.0-testing (rc9), and this syntax works here.

I don't suppose you have any more information in /var/log/daemon.log
or the xen logs (e.g., /var/log/xen/xend.log) ?



--
--
Best Regards

Minjia Zhang

Huazhong University of Science and Technology

Room 328,Dong 8 Building 8, No.1037 Luoyu Road, Wuhan 430074, P.R.C.

Tel: +86-159-2625-3437

Email:zhangninja@xxxxxxxxx
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.