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

Re: [ogfs-dev]Pool replacement strategy, EVMS and 2.5



On Monday 14 April 2003 07:04 pm, Cahill, Ben M wrote:
> > -----Original Message-----
> > From: Greg Freemyer [mailto:freemyer@NorcrossGroup.com]
> > Sent: Monday, April 14, 2003 4:28 PM
> > To: OpenGFS
> > Subject: re: [ogfs-dev]Pool replacement strategy, EVMS and 2.5
> >
> >
> > Ben,
> >
> > After we eliminate pool, we need to worry about supporting
> > dynamically adding a node.  right?
>
> Do you mean adding a new node *to the cluster description on the cidev*?
> Are there any current restrictions for the timing of adding nodes that are
> *already declared* in the cidev info?

Maybe somebody can add to the docs how exactly ogfs ties jounrals to a 
specific node. I don't think I've ever heard a really clear answer on this. I 
think it would help us make decisions going forward.

>
> I looked into this some when researching the mating of journals and cluster
> members, but a lot has slid out of my little brain . . . do we know why
> ogfs requires that *all potential* nodes be declared up front?

I think it is a historical thing from when ogfs didn't have the ability to add 
journals. Some of the docs floating around from Sistina still talk about that 
being on the TODO list. No doubt most people have read those, since most of 
the links used point straight at their sight, before they wised up and 
started moving things on a regular schedule.

>
> It sounds like a good thing to do . . . but I have no real opinion about
> sequencing of projects . . . once I get the basic external journal stuff
> together (I'm at code-complete stage right now, need to build and test with
> EVMS), I'll probably be thinking about the resizing issue, since that is
> part of what pool provided.

Have you done all the util changes as well? If not, or if you just want a 
second pair of eyes to peek at it, let me know.

>
> -- Ben --
>
> Opinions are mine, not Intel's
>
> > Once we have a command-line tool that works, that might be
> > another thing to use a EVMS FSIM module to invoke.
> >
> > (I realize it doesn't currently work, so this may be a
> > difficult step.)
> >
> > Greg
> > --
> > Greg Freemyer
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Opengfs-devel mailing list
> Opengfs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/opengfs-devel



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Opengfs-devel mailing list
Opengfs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opengfs-devel

[Kernel]     [Security]     [Bugtraq]     [Photo]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Clusters]     [Linux RAID]     [Yosemite Hiking]     [Linux Resources]

Powered by Linux