Good day
I'm using XCP 1.1 and have been testing out Intellicache. The VM I'm
experimenting with is a Windows 7 machine, running from an NFS SR. I have
enabled caching on the XCP host, pointing to a new EXT3 SR on an Intel SSD.
I then enabled caching of what is the VM's c: with this command:
xe vdi-param-set uuid=75eaac3e-d2ec-4f45-ac9f-67586f2d703c
allow-caching=true on-boot=persist
In testing, performance has been astonishing. Naturally, writes are limited
by NFS, but the raw speed boost we've seen is incredible. I'd dearly love to
leave Intellicache enabled, but I'm seeing predictable, regular corruption.
I have a VM protection policy enabled, set to perform a simple disk snapshot
each day at 23:45. After snapshotting, the cached vhd is, according to
Windows, corrupt. A read-only chkdsk reports hundreds of errors before
failing to continue in read-only mode.
I have now shut the VM down, disabled caching, and started it up. Windows
automatically scheduled a chkdsk on c: and is repairing as I type this. I
had hoped the "corruption", if that really is the case, would not have hit
the c:, but it has.
Searching online, I've found zero reported cases of this type of behaviour.
Does anyone here have an idea how and why corruption sneaks in?
Some notes:
I'm using xapi_version_override to emulate XenServer 5.6.100
The VM in question has dynamic memory allocated
VM protection policy is enabled, performing a daily, disk-based snapshot
--
View this message in context:
http://xen.1045712.n5.nabble.com/Intellicache-linked-to-vhd-corruption-tp4924147p4924147.html
Sent from the Xen - User mailing list archive at Nabble.com.
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home