I am trying to use virt-v2v to convert a Fedora 17 KVM guest to a RHEV guest. After creating a libvirt storage pool from my existing VG (uuhg!), I run the following command:
sudo virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Everything appeared to be working until the process (apparently) hung at 74%. ps shows that it is in a D+ state:
2991 pts/1 D+ 0:10 /usr/bin/perl /bin/virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Any ideas? Thanks!
On Fri, Nov 09, 2012 at 05:53:57PM -0600, Ian Pilcher wrote:
I am trying to use virt-v2v to convert a Fedora 17 KVM guest to a RHEV guest. After creating a libvirt storage pool from my existing VG (uuhg!), I run the following command:
sudo virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Everything appeared to be working until the process (apparently) hung at 74%. ps shows that it is in a D+ state:
2991 pts/1 D+ 0:10 /usr/bin/perl /bin/virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Any ideas? Thanks!
Odd, no I don't think I've seen anything like this.
The D state would indicate that it's probably lost an interrupt from the disk. Anything in 'dmesg'?
Rich.
On 11/11/12 21:04, Richard W.M. Jones wrote:
On Fri, Nov 09, 2012 at 05:53:57PM -0600, Ian Pilcher wrote:
I am trying to use virt-v2v to convert a Fedora 17 KVM guest to a RHEV guest. After creating a libvirt storage pool from my existing VG (uuhg!), I run the following command:
sudo virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Everything appeared to be working until the process (apparently) hung at 74%. ps shows that it is in a D+ state:
2991 pts/1 D+ 0:10 /usr/bin/perl /bin/virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Any ideas? Thanks!
Odd, no I don't think I've seen anything like this.
The D state would indicate that it's probably lost an interrupt from the disk. Anything in 'dmesg'?
You need to be looking in the kernel for the source of the D state. However, if you asked me to make a bet I'd say it's hanging trying to write to nfs. I'd investigate your network and your NFS server.
Incidentally, even if you get past this the conversion isn't going to work because v2v doesn't support F17 guests (yet). Upstream support for this hopefully coming fairly soon.
Matt
On 11/12/2012 05:35 AM, Matthew Booth wrote:
You need to be looking in the kernel for the source of the D state. However, if you asked me to make a bet I'd say it's hanging trying to write to nfs. I'd investigate your network and your NFS server.
Oddly enough the "NFS server" is the same system on which I was trying to run virt-v2v. Googling about gives me a sneaking feeling that this is a "bad thing"(TM).
Incidentally, even if you get past this the conversion isn't going to work because v2v doesn't support F17 guests (yet). Upstream support for this hopefully coming fairly soon.
Well, it was a RHEL 6 guest (on a F17 host). dd & netcat did the trick.
On 11/11/12 21:04, Richard W.M. Jones wrote:
On Fri, Nov 09, 2012 at 05:53:57PM -0600, Ian Pilcher wrote:
I am trying to use virt-v2v to convert a Fedora 17 KVM guest to a RHEV guest. After creating a libvirt storage pool from my existing VG (uuhg!), I run the following command:
sudo virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Everything appeared to be working until the process (apparently) hung at 74%. ps shows that it is in a D+ state:
2991 pts/1 D+ 0:10 /usr/bin/perl /bin/virt-v2v -o rhev -os 192.168.2.254:/srv/nfs --vmtype server -n rhevm rhev-ipa
Any ideas? Thanks!
Odd, no I don't think I've seen anything like this.
The D state would indicate that it's probably lost an interrupt from the disk. Anything in 'dmesg'?
You need to be looking in the kernel for the source of the D state. However, if you asked me to make a bet I'd say it's hanging trying to write to nfs. I'd investigate your network and your NFS server.
Incidentally, even if you get past this the conversion isn't going to work because v2v doesn't support F17 guests (yet). Upstream support for this hopefully coming fairly soon.
Matt
On 11/11/2012 03:04 PM, Richard W.M. Jones wrote:> Odd, no I don't think I've seen anything like this.
The D state would indicate that it's probably lost an interrupt from the disk. Anything in 'dmesg'?
Nothing stood out. (Going from memory as I gave up and did the migration with dd & and netcat. Had to reclaim the storage used by the export domain.)
But see my response to mbooth ...
-
Ian Pilcher -
Matthew Booth -
Richard W.M. Jones