Bug 348861

I did some more testing and it seems the message can easily reproduced by either 
using 'migrate' or 'save' against a running guest. In either case we will see
 the message 

libvir: Xen Store error : out of memory
libvir: error : invalid argument in __virGetDomain 

A quick strace of the virt-manager process (extract from 10k trace)
Note vBlade1 is the guest which is being "shutdown" as part of a 
"virsh save vBlade1 vBlade1.sav" command

rt_sigaction(SIGPIPE, {SIG_IGN}, {SIG_IGN}, 8) = 0
write(16, "\2\0\0\0\0\0\0\0\0\0\0\0\26\0\0\0", 16) = 16
write(16, "/local/domain/45/name\0", 22) = 22
read(16, "\2\0\0\0\0\0\0\0\0\0\0\0\5\0\0\0", 16) = 16
read(16, "WinXP", 5)                    = 5
rt_sigaction(SIGPIPE, {SIG_IGN}, NULL, 8) = 0
rt_sigaction(SIGPIPE, {SIG_IGN}, {SIG_IGN}, 8) = 0
write(16, "\2\0\0\0\0\0\0\0\0\0\0\0\26\0\0\0", 16) = 16
write(16, "/local/domain/46/name\0", 22) = 22
read(16, "\2\0\0\0\0\0\0\0\0\0\0\0\7\0\0\0", 16) = 16
read(16, "vBlade1", 7)                  = 7
rt_sigaction(SIGPIPE, {SIG_IGN}, NULL, 8) = 0
rt_sigaction(SIGPIPE, {SIG_IGN}, {SIG_IGN}, 8) = 0
write(16, "\n\0\0\0\0\0\0\0\0\0\0\0\3\0\0\0", 16) = 16
write(16, "46\0", 3)                    = 3
read(16, "\n\0\0\0\0\0\0\0\0\0\0\0\21\0\0\0", 16) = 16
read(16, "/local/domain/46\0", 17)      = 17
rt_sigaction(SIGPIPE, {SIG_IGN}, NULL, 8) = 0
write(2, "libvir: error : invalid argument"..., 51) = 51
write(2, "libvir: Xen Store error : out of"..., 40) = 40
time(NULL)                              = 1193151043

Created attachment 292113 [details]
xsat xen config

the config for my domU

Created attachment 292115 [details]
my xend config

my xend config

my comments again: (got confused with the attachments :( )

i have the same error messages. The odd thing is that live migration worked.

I use :
rhel 5.1
xen-3.0.3-41.el5
libvirt-0.2.3-9.el5

the migration was started with the xm command :
xm migrate xsat node2

Problem not reproduceable after fixing xsat xen config and restarting the nodes.
I removed the r from the disc line, so it reads w now:

before fix:
disk = [ 'phy:/dev/mapper/VGSAN-master_root,xvda,wr',
'phy:/dev/mapper/VGSAN-master_satellite,xvdb,w' ]

after fix:
disk = [ 'phy:/dev/mapper/VGSAN-master_root,xvda,w',
'phy:/dev/mapper/VGSAN-master_satellite,xvdb,w' ]

Okay, this message is being printed by libvirt, so is not specific to virt-manager. 'virsh save' would cause the same result.

However, with the libvirt version queued up for 5.4, I can no longer reproduce this. So we will probably get this for free. It could even be fixed on stock 5.3, but I haven't tested.

Reassigning to libvirt.

Yes that looks fixed to me with the current rebase, at least I wasn't
able to reproduce the problem anymore with 0.6.2 built for RHEL-5:

[root@test2 ~]# virsh list
 Id Name                 State
----------------------------------
  0 Domain-0             running
  4 migr5                idle

[root@test2 ~]# virsh save migr5 /tmp/migr5.save
Domain migr5 saved to /tmp/migr5.save

[root@test2 ~]# rpm -q libvirt
libvirt-0.6.2-1.el5
[root@test2 ~]# virsh restore /tmp/migr5.save
Domain restored from /tmp/migr5.save

[root@test2 ~]# virsh list
 Id Name                 State
----------------------------------
  0 Domain-0             running
  5 migr5                idle

[root@test2 ~]# 

Daniel

Test on libvirt 0.6.3-3, the FV guest is running with 2G memory and both the dom-0 systems have 16G physical memory on rhel-5.4.

virt-manager error info:

Error migrating domain: POST operation failed: xend_post: error from xen daemon: (xend.err '/usr/lib64/xen/bin/xc_save 17 6 0 0 5 failed')

Traceback (most recent call last):
  File "/usr/share/virt-manager/virtManager/engine.py", line 561, in migrate_domain
    vm.migrate(destconn)
  File "/usr/share/virt-manager/virtManager/domain.py", line 1378, in migrate
    self.vm.migrate(self.connection.vmm, flags, None, dictcon.get_short_hostname(), 0)
  File "/usr/lib64/python2.4/site-packages/libvirt.py", line 378, in migrate
    if ret is None:raise libvirtError('virDomainMigrate() failed', dom=self)
libvirtError: POST operation failed: xend_post: error from xen daemon: (xend.err '/usr/lib64/xen/bin/xc_save 17 6 0 0 5 failed')

virsh error info:

[root@intel-5130-32-1 ~]# virsh list --all
 Id Name                 State
----------------------------------
  0 Domain-0             running
  1 foo                  idle
[root@intel-5130-32-1 ~]# virsh migrate foo xen+ssh://10.66.83.192
root.83.192's password: 
error: POST operation failed: xend_post: error from xen daemon: (xend.err "can't connect: Name or service not known")

comment #11 looks like a migration errror. But as #1 pointed that
bug could be reproduced with the save command, and in #8 it disapeared.
Now if migration fails that could be related to a variety of other
reasons, let's not mix things, we need to check migration again, but
I think it should be a different bug, especially with the error reported

  "Name or service not known"

is clearly something completely different than the original bug,
which to me is still in MODIFIED state.

Daniel

Actually this bug was already fixed, but still has the error "Name or service not known", for this issue will open a new bug later.


[root@intel-5130-32-1 ~]# virsh start demo
Domain demo started

[root@intel-5130-32-1 ~]# virsh migrate demo xen+ssh://10.66.83.192
root.83.192's password: 
error: POST operation failed: xend_post: error from xen daemon: (xend.err "can't connect: Name or service not known")

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHEA-2009-1269.html