Xen users - Jul 2010 - XCP 0.5.0 release, upgrade bugs summary 0.1.1 to 0.5.0

Ok, finally got the cloud straightened out and flying again (whew). 
Here''s my summary of upgrading bugs and caveats:

Attempting to list all VMs from xsconsole produces "This feature is not 
available in pools with more than 100 virtual machines." when there are 
less than 100 VMs, even with a fully homogenous 0.5.0 pool. Xen Cloud 
Control System is not affected by this bug.

VMs with 0.1.1 pv drivers show that the pv drivers are up to date from 
the 0.5.0 pool master. Obviously incorrect. XCCS is affected by this 
bug, workaround will be in place for XCCS 0.5 release.

Upgrading the guest utilities to 0.5.0 by using installing the rpm 
directly with rpm requires using --force to override errors that 
complain about conflicts with the 0.1.1 guest utilities. This does not 
occur if you use the install.sh script to install them.

With the pool in a partially upgraded state, 0.5 pool master and 0.1.1 
slaves, you can''t migrate a VM to the pool master to clear a slave for 
upgrading. Any attempt to migrate off a slave, shut the VMs down, or 
even shut the slave down produces "Failed: Internal error: Not_found".
Fix was to dump the slaves the hard way and restart the VMs on the 
upgraded pool master.

-- 
Vern Burke

SwiftWater Telecom
http://www.swiftwatertel.com
ISP/CLEC Engineering Services
Data Center Services
Remote Backup Services

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi Vern,

Thanks for the summary. Could you put these into the bugzilla?

The worst one sounds like the Not_found one at the end -- I don''t
suppose you still have the logs lying around? I''d quite like to
understand that one in more detail. The rest I can probably work out without
logs.

Cheers,
Dave
> -----Original Message-----
> From: Vern Burke [mailto:vburke@skow.net]
> Sent: 07 July 2010 18:39
> To: xen-users@lists.xensource.com
> Cc: Dave Scott
> Subject: XCP 0.5.0 release, upgrade bugs summary 0.1.1 to 0.5.0
> 
> Ok, finally got the cloud straightened out and flying again (whew).
> Here''s my summary of upgrading bugs and caveats:
> 
> Attempting to list all VMs from xsconsole produces "This feature is
not
> available in pools with more than 100 virtual machines." when there
are
> less than 100 VMs, even with a fully homogenous 0.5.0 pool. Xen Cloud
> Control System is not affected by this bug.
> 
> VMs with 0.1.1 pv drivers show that the pv drivers are up to date from
> the 0.5.0 pool master. Obviously incorrect. XCCS is affected by this
> bug, workaround will be in place for XCCS 0.5 release.
> 
> Upgrading the guest utilities to 0.5.0 by using installing the rpm
> directly with rpm requires using --force to override errors that
> complain about conflicts with the 0.1.1 guest utilities. This does not
> occur if you use the install.sh script to install them.
> 
> With the pool in a partially upgraded state, 0.5 pool master and 0.1.1
> slaves, you can''t migrate a VM to the pool master to clear a slave
for
> upgrading. Any attempt to migrate off a slave, shut the VMs down, or
> even shut the slave down produces "Failed: Internal error:
Not_found".
> Fix was to dump the slaves the hard way and restart the VMs on the
> upgraded pool master.
> 
> --
> Vern Burke
> 
> SwiftWater Telecom
> http://www.swiftwatertel.com
> ISP/CLEC Engineering Services
> Data Center Services
> Remote Backup Services
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Bug exists at: 

XCP 0.1.1
XCP 0.5RC2,RC3
XCP 0.5

Bug description:

memory-actual of guest machine (PV) can not surpass value of
memory-dynamic-max at machine starttime, even if this value
(memory-dynamic-max have been raised after boot).

Steps to reproduce:

(on any PV guest VM)

xe vm-shutdown vm=test
xe vm-memory-static-range-set min=100MiB max=2GiB vm=test
xe vm-memory-dynamic-range-set min=100MiB max=500MiB vm=test
xe vm-start vm=test
(after booting)

name-label ( RW)            : test
         memory-actual ( RO): 524288000
     memory-static-max ( RW): 2147483648
    memory-dynamic-max ( RW): 524288000 (* let call it boot_max)
    memory-dynamic-min ( RW): 104857600
     memory-static-min ( RW): 104857600

xe vm-memory-dynamic-range set max=200MiB min=200MiB vm=test
(here all ok)
           power-state ( RO): running
         memory-actual ( RO): 209715200
     memory-static-max ( RW): 2147483648
    memory-dynamic-max ( RW): 209715200
    memory-dynamic-min ( RW): 209715200
     memory-static-min ( RW): 104857600


and here is a bug:
name-label ( RW)            : xv-acc1
           power-state ( RO): running
         memory-actual ( RO): 524288000
     memory-static-max ( RW): 2147483648
    memory-dynamic-max ( RW): 838860800
    memory-dynamic-min ( RW): 838860800
     memory-static-min ( RW): 104857600

we changed an memory-dynamic-max up to 800MiB, but memory-actual does
not overcome boot_max value, mentioned upper).

I check situation with dynamic-min but it work as expected - if I move
dynamic_min & dynamic_max below value of dynamic_min at boot time, it
move memory_actual  below boot time dynamic_min.





_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users