Re: [ogfs-dev]RFC: Reclaim the excess freemeta blocks

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



OK. I will send the initial patch in few days.

Thanks,
Stan

On Fri, 2003-06-13 at 19:49, Dominik Vogt wrote:
> On Fri, Jun 13, 2003 at 07:22:44PM +0800, Stanley Wang wrote:
> > It sounds good.
> > In fact, I want to do the reclaim work in a user space tool originally
> > too. It's safer and easier way. But this way means that you must umount
> > the SAN-disks from every nodes. I don't know whether it is acceptable.
> > Hence I choice to implement the reclaiming function in ogfs_inoded. And
> > we could reclaim excess freemeta blocks without influencing the normal
> > work of ogfs. And of course this solution should cause performance
> > downgrad. The key point is whether it is worthy to do so.
> >  
> > And another choice: the user space tool is just a ioctl or something
> > alike, it is the same as the implementation in ogfs_inoded() execpt for
> > less performance downgrade. The reclaiming action is only executed when
> > we want. How do you think about this solution ?
> 
> Sounds good as long we don't let it reclaim blocks automatically.
> By the way, the same thing would be useful for freeing inode
> blocks.
> 
> > On Fri, 2003-06-13 at 15:49, Dominik Vogt wrote:
> > > On Fri, Jun 13, 2003 at 11:47:18AM +0800, Stanley Wang wrote:
> > > > Hi, Folks
> > > > I found that the ogfs couldn't reclaim the excess freemeta blocks to the
> > > > freedata blocks When I read the On-Disk layout doc. Hence I think that
> > > > we should add a cleaner to resolve this issue.
> > > > 
> > > > My plan is that ogfs_inoded should be responsible for scaning all rgrp
> > > > and reclaim the excess freemeta. It go though the rgrp list and clean
> > > > the rgrps one by one when waked up. 
> > > > 
> > > > Any comments or suggestinos?
> > > 
> > > I suggest to do it as a standalone tool for the beginning.  Ogfs
> > > tries to cluster its meta blocks and simply converting them back
> > > to data blocks once a high watermark is reached may have
> > > catastrophic effects on keeping data local on the disk.
> > > 
> > > > I will send out the patch in few days if you feel good for this RFC.
> 
> Bye
> 
> Dominik ^_^  ^_^
> 
> 
> -------------------------------------------------------
> This SF.NET email is sponsored by: eBay
> Great deals on office technology -- on eBay now! Click here:
> http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5
> _______________________________________________
> Opengfs-devel mailing list
> Opengfs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/opengfs-devel
-- 
Opinions expressed are those of the author and do not represent Intel
Corporation



-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5
_______________________________________________
Opengfs-devel mailing list
Opengfs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opengfs-devel

[Index of Archives]     [Kernel]     [Security]     [Bugtraq]     [MIPS Linux]     [ARM Linux]     [Linux Clusters]     [Linux RAID]     [Yosemite Hiking]
  Powered by Linux