Sorry, my fault, attention is elsewhere ..
I actually use "writethrough" rather than "writeback" .. so with the exception
of the quote from the
mailing list, where I said "writethrough", please read "writeback" ...
apologies.
I "used" to use writeback, but then read a few horror stories and tried
writethrough .. and when used
with vdc-store, writethrough .vs. writeback makes almost no difference anyway
as vdc-store buffers the
output "and" is migration safe ...
I've done (literally) thousands of migrations using VM's under IO load with
writethrough cache (damn!
I just typed writeback again and had to correct! losing my mind!) while
testing with ON, never had a
problem or corruption - so for me writethrough is certainly safe enough to make
me want to use the
sharable flag .. :)
--
Gareth Bult
“The odds of hitting your target go up dramatically when you aim at it.”
See the status of my current project at http://vdc-store.com
----- Original Message -----
From: "Stefan Kooman" <ste...@bit.nl>
To: "Gareth Bult" <gar...@linux.co.uk>
Cc: "Javier Fontan" <jfon...@opennebula.org>, users@lists.opennebula.org
Sent: Thursday, 30 January, 2014 2:59:33 PM
Subject: Re: [one-users] Setting the "SHAREABLE" attribute
Quoting Gareth Bult (gar...@linux.co.uk):
> Mmm, there has been some discussion around this ..
>
> Technically using cache=writeback "should" be safe for migration.
> (which is the method I use)
>
> So .. I need "cache=writeback" and "sharable" for migration to happen.
> (the performance hit with cache=off is an unacceptable performance hit, at
> least for me ..)
>
> From the QEMU lists, 2012;
> "In short, if you're using a recent kernel with ext3 or ext4, cache=writeback
> is absolutely safe. If you're using an older version of ext3, cache=writeback
> is still safe but ext3 itself isn't. cache=writeback can make the situation
> worse."
Hmm, the guys at big blue think otherwise [1]:
With caching set to writeback mode, both the host page cache and the
disk write cache are enabled for the guest. Because of this, the I/O
performance for applications running in the guest is good, but the data
is not protected in a power failure. As a result, this caching mode is
recommended only for temporary data where potential data loss is not a
concern.
Gr. Stefan
[1]:
http://pic.dhe.ibm.com/infocenter/lnxinfo/v3r0m0/index.jsp?topic=%2Fliaat%2Fliaatbpkvmasynchio.htm
--
| BIT BV http://www.bit.nl/ Kamer van Koophandel 09090351
| GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl
_______________________________________________
Users mailing list
Users@lists.opennebula.org
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
