<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7638.1">
<TITLE>RE: [Gluster-users] Distrbute AFR Storage</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>Is there a way to use a round-robin or something, as opposed to hash?<BR>
<BR>
The problem with distribute is that it does not know the disk has run out of space, until it's starting copying and run out of space.<BR>
Can this be avoided?<BR>
<BR>
-----Original Message-----<BR>
From: seandavi@gmail.com on behalf of Sean Davis<BR>
Sent: Sat 3/28/2009 9:51 PM<BR>
To: Simon Liang<BR>
Cc: Vikas Gorur; gluster-users<BR>
Subject: Re: [Gluster-users] Distrbute AFR Storage<BR>
<BR>
<BR>
<BR>
2009/3/28 Simon Liang <simonl@bigair.net.au><BR>
<BR>
<BR>
Well currently I've got 4 servers set up.<BR>
|- ServerA -AFR-> ServerB<BR>
Client -|<BR>
|- ServerC -AFR-> ServerD<BR>
<BR>
I keep copying the same 1GB file to the storage, but it always writes to ServerC no matter what I do. Is there anything I can do to fix this? Add scheduler?<BR>
<BR>
<BR>
<BR>
My understanding is that distribute uses a hash to do the distribution. So, if you copy the same file, you will get the same result every time; that is, it will go to the same server every time. Hashing is deterministic in that sense. Vikas or someone else may clarify if I have things wrong here....<BR>
<BR>
Sean<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
-----Original Message-----<BR>
From: vikasgp@gmail.com on behalf of Vikas Gorur<BR>
<BR>
Sent: Sat 3/28/2009 3:53 PM<BR>
To: Simon Liang<BR>
Cc: gluster-users<BR>
Subject: Re: [Gluster-users] Distrbute AFR Storage<BR>
<BR>
<BR>
2009/3/28 Simon Liang <simonl@bigair.net.au>:<BR>
> How can I configure it so it writes it evenly to both servers, keeping the<BR>
> free disk space even?<BR>
<BR>
Distribute by default will distribute _files_ evenly among its<BR>
subvolumes. However, if one of the files is larger than the disk space<BR>
available on the node it gets scheduled to, then distribute can do<BR>
nothing. If you regularly want to store big files which might not fit<BR>
on one of your subvolumes, you might want to try the 'stripe'<BR>
translator.<BR>
<BR>
Vikas<BR>
--<BR>
Engineer - Z Research<BR>
<A HREF="http://gluster.com/">http://gluster.com/</A><BR>
<BR>
<BR>
No virus found in this incoming message.<BR>
Checked by AVG - www.avg.com<BR>
<BR>
Version: 8.0.238 / Virus Database: 270.11.29/2024 - Release Date: 03/27/09 18:51:00<BR>
<BR>
<BR>
<BR>
<BR>
_______________________________________________<BR>
Gluster-users mailing list<BR>
Gluster-users@gluster.org<BR>
<A HREF="http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users">http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users</A><BR>
<BR>
<BR>
<BR>
<BR>
No virus found in this incoming message.<BR>
Checked by AVG - www.avg.com<BR>
Version: 8.0.238 / Virus Database: 270.11.29/2024 - Release Date: 03/27/09 18:51:00<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>