[Users] ploop and trim/ discard support

Corin Langosch corin.langosch at netskin.com
Wed Sep 12 17:38:59 EDT 2012


On 12.09.2012 at 21:34 +0200, Kir Kolyshkin <kir at openvz.org> wrote:
>
> On Sep 12, 2012 11:20 PM, "Corin Langosch" <info at corinlangosch.com 
> <mailto:info at corinlangosch.com>> wrote:
> >
> > Hi,
> >
> > it seems that when creating & deleting files inside a ploop backed 
> container the backing image file doesn't shrink. So the image file is 
> still consuming around 400 GB, while in the container df shows only 
> 100 GB in use.
> >
> > ll on host: -rw------- 1 root root 384G Sep 12 21:06 root.hdd
> > df on container: /dev/ploop1           493G  103G  390G  21% /
>
> We have online shrink. It's "ploop balloon discard" or just "vzctl 
> compact". You'd better have ploop and vzctl from git (both are really 
> close to be released).
>

The system's still working heavily but the image is already reduced by 
around 60 GB so it seems to work well. :)

Would you suggest putting some script into crontab to compact all ploop 
images ex. once a week? Or probably do something like ctid%7 and compact 
a few of them every night?

BTW: the man of vzctl contains the "compact" command. But the usage 
output of vzctl doesn't. Shall I file a bug report for it?

BTW 2: I just read about the ploop ballooning technique in the wiki. Am 
I correct in that it works quite similar to sdelete on windows ( 
http://pastie.org/4710364 or 
http://technet.microsoft.com/en-us/sysinternals/bb897443.aspx)?

Thanks!

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://openvz.org/pipermail/users/attachments/20120912/6f0a4155/attachment.html


More information about the Users mailing list