<!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 6.5.7235.2">
<TITLE>RE: Lustre-discuss Digest, Vol 26, Issue 25</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<BR>
<P><FONT SIZE=2>Can't we build patched lustre server for 2.6.20.15vanilla kernel ?<BR>
<BR>
Regards,<BR>
Ashok Bharat<BR>
<BR>
-----Original Message-----<BR>
From: lustre-discuss-bounces@lists.lustre.org on behalf of lustre-discuss-request@lists.lustre.org<BR>
Sent: Mon 3/10/2008 4:58 PM<BR>
To: lustre-discuss@lists.lustre.org<BR>
Subject: Lustre-discuss Digest, Vol 26, Issue 25<BR>
<BR>
Send Lustre-discuss mailing list submissions to<BR>
lustre-discuss@lists.lustre.org<BR>
<BR>
To subscribe or unsubscribe via the World Wide Web, visit<BR>
<A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
or, via email, send a message with subject or body 'help' to<BR>
lustre-discuss-request@lists.lustre.org<BR>
<BR>
You can reach the person managing the list at<BR>
lustre-discuss-owner@lists.lustre.org<BR>
<BR>
When replying, please edit your Subject line so it is more specific<BR>
than "Re: Contents of Lustre-discuss digest..."<BR>
<BR>
<BR>
Today's Topics:<BR>
<BR>
1. Re: Kernel Patch Management (Andreas Dilger)<BR>
2. Re: yet another lustre error (Alex Lyashkov)<BR>
3. Re: lustre and small files overhead (Joe Barjo)<BR>
<BR>
<BR>
----------------------------------------------------------------------<BR>
<BR>
Message: 1<BR>
Date: Mon, 10 Mar 2008 00:36:12 -0600<BR>
From: Andreas Dilger <adilger@sun.com><BR>
Subject: Re: [Lustre-discuss] Kernel Patch Management<BR>
To: ashok bharat bayana <ashok.bharat.bayana@iiitb.ac.in><BR>
Cc: lustre-discuss@lists.lustre.org<BR>
Message-ID: <20080310063612.GF5851@webber.adilger.int><BR>
Content-Type: text/plain; charset=us-ascii<BR>
<BR>
On Mar 10, 2008 11:21 +0530, ashok bharat bayana wrote:<BR>
> I'm trying to build lustre(1.6.4.2) version on 2.6.20.15 kernel on an x86 based machine.I'm using Ubuntu 7.10 flavour.<BR>
><BR>
> While applying patches to the kernel in the series which_patch file,under the supported kernels, vanilla-2.6.20.15.series is not menioned but vanilla-2.6.18-.series is mentioned.<BR>
><BR>
> So, can I proceed applying the series file of 2.6.18 version on a 2.6.20.15 version.<BR>
<BR>
No, making Lustre kernel patches is not so easy. You should instead use<BR>
a supported kernel version for the server, and use patchless Lustre clients<BR>
for any systems which cannot run on a supported kernel.<BR>
<BR>
> -----Original Message-----<BR>
> From: lustre-discuss-bounces@lists.lustre.org on behalf of lustre-discuss-request@lists.lustre.org<BR>
> Sent: Fri 2/29/2008 12:17 PM<BR>
> To: lustre-discuss@lists.lustre.org<BR>
> Subject: Lustre-discuss Digest, Vol 25, Issue 58<BR>
> <BR>
> Send Lustre-discuss mailing list submissions to<BR>
> lustre-discuss@lists.lustre.org<BR>
><BR>
> To subscribe or unsubscribe via the World Wide Web, visit<BR>
> <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> or, via email, send a message with subject or body 'help' to<BR>
> lustre-discuss-request@lists.lustre.org<BR>
><BR>
> You can reach the person managing the list at<BR>
> lustre-discuss-owner@lists.lustre.org<BR>
><BR>
> When replying, please edit your Subject line so it is more specific<BR>
> than "Re: Contents of Lustre-discuss digest..."<BR>
><BR>
><BR>
> Today's Topics:<BR>
><BR>
> 1. Re: Configuration of lustre FS on single node<BR>
> (ashok bharat bayana)<BR>
> 2. Re: Configuration of lustre FS on single node (Amit Sharma)<BR>
> 3. Re: Configuration of lustre FS on single node<BR>
> (ashok bharat bayana)<BR>
><BR>
><BR>
> ----------------------------------------------------------------------<BR>
><BR>
> Message: 1<BR>
> Date: Fri, 29 Feb 2008 12:07:18 +0530<BR>
> From: "ashok bharat bayana" <ashok.bharat.bayana@iiitb.ac.in><BR>
> Subject: Re: [Lustre-discuss] Configuration of lustre FS on single<BR>
> node<BR>
> To: <lustre-discuss@lists.lustre.org>,<BR>
> <lustre-discuss@lists.lustre.org><BR>
> Message-ID: <8626C1B7EB748940BCDD7596134632BE850229@jal.iiitb.ac.in><BR>
> Content-Type: text/plain; charset="iso-8859-1"<BR>
><BR>
><BR>
> Hello,<BR>
> I successfully build lustre(1.6.4.2) on my system for a patchless client But I dont know how to proceed in configuring the file system.<BR>
><BR>
> I'm trying to build all MDS,client and MDT's on a single node.<BR>
> >From the tutorials I came to know that<BR>
><BR>
> First, create an MDT for the "spfs" file system that uses the /dev/sda disk. This MDT also acts as the MGS for the site.<BR>
><BR>
> $ mkfs.lustre --fsname spfs --mdt --mgs /dev/sda<BR>
><BR>
> But by triggering this command my system crashed in which the whole data is lost and again I need to install the OS.<BR>
> I want help in proceeding of mounting a lustre file system.<BR>
><BR>
> Thanks and Regards,<BR>
> Ashok Bharat<BR>
><BR>
> -----Original Message-----<BR>
> From: lustre-discuss-bounces@lists.lustre.org on behalf of lustre-discuss-request@lists.lustre.org<BR>
> Sent: Wed 2/27/2008 10:30 PM<BR>
> To: lustre-discuss@lists.lustre.org<BR>
> Subject: Lustre-discuss Digest, Vol 25, Issue 56<BR>
> <BR>
> Send Lustre-discuss mailing list submissions to<BR>
> lustre-discuss@lists.lustre.org<BR>
><BR>
> To subscribe or unsubscribe via the World Wide Web, visit<BR>
> <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> or, via email, send a message with subject or body 'help' to<BR>
> lustre-discuss-request@lists.lustre.org<BR>
><BR>
> You can reach the person managing the list at<BR>
> lustre-discuss-owner@lists.lustre.org<BR>
><BR>
> When replying, please edit your Subject line so it is more specific<BR>
> than "Re: Contents of Lustre-discuss digest..."<BR>
><BR>
><BR>
> Today's Topics:<BR>
><BR>
> 1. Multiple NICs per OST (Joshua Bower-Cooley)<BR>
> 2. obd error on MGS/MDT node (Jeremy Mann)<BR>
><BR>
><BR>
> ----------------------------------------------------------------------<BR>
><BR>
> Message: 1<BR>
> Date: Tue, 26 Feb 2008 11:55:19 -0800<BR>
> From: Joshua Bower-Cooley <jbowercooley@lcogt.net><BR>
> Subject: [Lustre-discuss] Multiple NICs per OST<BR>
> To: lustre-discuss@lists.lustre.org<BR>
> Message-ID: <200802261155.19907.jbowercooley@lcogt.net><BR>
> Content-Type: text/plain; charset="utf-8"<BR>
><BR>
> Hi-<BR>
> I've had lustre running happily for some time over a single 10g ethernet NIC<BR>
> per node. After switching to dual 10g and creating a new filesystem<BR>
> (1.6.4.2), I'm seeing nothing but keep-alive packets with bad checksums.<BR>
><BR>
> What is the current "correct" way to do this now? The manual suggest not using<BR>
> bonding, but several list postings now reccommend it. Without bonding, do I<BR>
> need to have my 2 switches stacked, or will Lustre recognize the division in<BR>
> my subnet?<BR>
><BR>
> LNET module options I've tried are:<BR>
> 1) networks="tcp0(eth2,eth3)"<BR>
> 2) ip2nets="tcp(eth2,eth3); tcp(eth2) 10.9.[1-4].*; tcp(eth3) 10.9.[5-8].*;"<BR>
> and many other variations<BR>
><BR>
> All other services are functional on both interfaces.<BR>
><BR>
> Thanks in advance,<BR>
> J Bower-Cooley<BR>
> Scientific Systems Engineer<BR>
> Las Cumbres Observatory Global Telescope<BR>
><BR>
><BR>
> ------------------------------<BR>
><BR>
> Message: 2<BR>
> Date: Wed, 27 Feb 2008 09:50:08 -0600 (CST)<BR>
> From: "Jeremy Mann" <jeremy@biochem.uthscsa.edu><BR>
> Subject: [Lustre-discuss] obd error on MGS/MDT node<BR>
> To: lustre-discuss@lists.lustre.org<BR>
> Message-ID:<BR>
> <62901.24.173.62.147.1204127408.squirrel@biochem.uthscsa.edu><BR>
> Content-Type: text/plain;charset=iso-8859-1<BR>
><BR>
> Today we are starting to see obd_change_cbdata errors on our MGS/MDT node.<BR>
> What does this error mean?<BR>
><BR>
> Version of lustre and kernel is:<BR>
><BR>
> 2.6.9-42.0.10.EL_lustre-1.6.0.1smp X86_64<BR>
><BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> previous similar message<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> obd_change_cbdata: NULL export<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> previous similar message<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> obd_change_cbdata: NULL export<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 2<BR>
> previous similar messages<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> obd_change_cbdata: NULL export<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 4<BR>
> previous similar messages<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> obd_change_cbdata: NULL export<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 5<BR>
> previous similar messages<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> obd_change_cbdata: NULL export<BR>
> LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> previous similar message<BR>
><BR>
> --<BR>
> Jeremy Mann<BR>
> jeremy@biochem.uthscsa.edu<BR>
><BR>
> University of Texas Health Science Center<BR>
> Bioinformatics Core Facility<BR>
> <A HREF="http://www.bioinformatics.uthscsa.edu">http://www.bioinformatics.uthscsa.edu</A><BR>
> Phone: (210) 567-2672<BR>
><BR>
><BR>
><BR>
> ------------------------------<BR>
><BR>
> _______________________________________________<BR>
> Lustre-discuss mailing list<BR>
> Lustre-discuss@lists.lustre.org<BR>
> <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
><BR>
><BR>
> End of Lustre-discuss Digest, Vol 25, Issue 56<BR>
> **********************************************<BR>
><BR>
> -------------- next part --------------<BR>
> An HTML attachment was scrubbed...<BR>
> URL: <A HREF="http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080229/887613dd/attachment-0001.html">http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080229/887613dd/attachment-0001.html</A><BR>
><BR>
> ------------------------------<BR>
><BR>
> Message: 2<BR>
> Date: Fri, 29 Feb 2008 12:03:09 +0530<BR>
> From: Amit Sharma <Amit.Sharma@Sun.COM><BR>
> Subject: Re: [Lustre-discuss] Configuration of lustre FS on single<BR>
> node<BR>
> To: ashok bharat bayana <ashok.bharat.bayana@iiitb.ac.in><BR>
> Cc: lustre-discuss@lists.lustre.org<BR>
> Message-ID: <47C7A725.5050309@sun.com><BR>
> Content-Type: text/plain; format=flowed; charset=ISO-8859-1<BR>
><BR>
> Ashok,<BR>
><BR>
> Could you let us know what are the modules that you have installed on<BR>
> the system. The linux version on your system. And the steps that you<BR>
> have done so far.<BR>
><BR>
> thanks,<BR>
> Amit<BR>
> ashok bharat bayana wrote:<BR>
> ><BR>
> > Hello,<BR>
> > I successfully build lustre(1.6.4.2) on my system for a patchless client<BR>
> > But I dont know how to proceed in configuring the file system.<BR>
> ><BR>
> > I'm trying to build all MDS,client and MDT's on a single node.<BR>
> > >From the tutorials I came to know that<BR>
> ><BR>
> > First, create an MDT for the "spfs" file system that uses the /dev/sda<BR>
> > disk. This MDT also acts as the MGS for the site.<BR>
> ><BR>
> > $ mkfs.lustre --fsname spfs --mdt --mgs /dev/sda<BR>
> ><BR>
> > But by triggering this command my system crashed in which the whole data<BR>
> > is lost and again I need to install the OS.<BR>
> > I want help in proceeding of mounting a lustre file system.<BR>
> ><BR>
> > Thanks and Regards,<BR>
> > Ashok Bharat<BR>
> ><BR>
> > -----Original Message-----<BR>
> > From: lustre-discuss-bounces@lists.lustre.org on behalf of<BR>
> > lustre-discuss-request@lists.lustre.org<BR>
> > Sent: Wed 2/27/2008 10:30 PM<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Subject: Lustre-discuss Digest, Vol 25, Issue 56<BR>
> ><BR>
> > Send Lustre-discuss mailing list submissions to<BR>
> > lustre-discuss@lists.lustre.org<BR>
> ><BR>
> > To subscribe or unsubscribe via the World Wide Web, visit<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> > or, via email, send a message with subject or body 'help' to<BR>
> > lustre-discuss-request@lists.lustre.org<BR>
> ><BR>
> > You can reach the person managing the list at<BR>
> > lustre-discuss-owner@lists.lustre.org<BR>
> ><BR>
> > When replying, please edit your Subject line so it is more specific<BR>
> > than "Re: Contents of Lustre-discuss digest..."<BR>
> ><BR>
> ><BR>
> > Today's Topics:<BR>
> ><BR>
> > 1. Multiple NICs per OST (Joshua Bower-Cooley)<BR>
> > 2. obd error on MGS/MDT node (Jeremy Mann)<BR>
> ><BR>
> ><BR>
> > ----------------------------------------------------------------------<BR>
> ><BR>
> > Message: 1<BR>
> > Date: Tue, 26 Feb 2008 11:55:19 -0800<BR>
> > From: Joshua Bower-Cooley <jbowercooley@lcogt.net><BR>
> > Subject: [Lustre-discuss] Multiple NICs per OST<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Message-ID: <200802261155.19907.jbowercooley@lcogt.net><BR>
> > Content-Type: text/plain; charset="utf-8"<BR>
> ><BR>
> > Hi-<BR>
> > I've had lustre running happily for some time over a single 10g ethernet NIC<BR>
> > per node. After switching to dual 10g and creating a new filesystem<BR>
> > (1.6.4.2), I'm seeing nothing but keep-alive packets with bad checksums.<BR>
> ><BR>
> > What is the current "correct" way to do this now? The manual suggest not<BR>
> > using<BR>
> > bonding, but several list postings now reccommend it. Without bonding, do I<BR>
> > need to have my 2 switches stacked, or will Lustre recognize the division in<BR>
> > my subnet?<BR>
> ><BR>
> > LNET module options I've tried are:<BR>
> > 1) networks="tcp0(eth2,eth3)"<BR>
> > 2) ip2nets="tcp(eth2,eth3); tcp(eth2) 10.9.[1-4].*; tcp(eth3) 10.9.[5-8].*;"<BR>
> > and many other variations<BR>
> ><BR>
> > All other services are functional on both interfaces.<BR>
> ><BR>
> > Thanks in advance,<BR>
> > J Bower-Cooley<BR>
> > Scientific Systems Engineer<BR>
> > Las Cumbres Observatory Global Telescope<BR>
> ><BR>
> ><BR>
> > ------------------------------<BR>
> ><BR>
> > Message: 2<BR>
> > Date: Wed, 27 Feb 2008 09:50:08 -0600 (CST)<BR>
> > From: "Jeremy Mann" <jeremy@biochem.uthscsa.edu><BR>
> > Subject: [Lustre-discuss] obd error on MGS/MDT node<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Message-ID:<BR>
> > <62901.24.173.62.147.1204127408.squirrel@biochem.uthscsa.edu><BR>
> > Content-Type: text/plain;charset=iso-8859-1<BR>
> ><BR>
> > Today we are starting to see obd_change_cbdata errors on our MGS/MDT node.<BR>
> > What does this error mean?<BR>
> ><BR>
> > Version of lustre and kernel is:<BR>
> ><BR>
> > 2.6.9-42.0.10.EL_lustre-1.6.0.1smp X86_64<BR>
> ><BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 2<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 4<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 5<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> ><BR>
> > --<BR>
> > Jeremy Mann<BR>
> > jeremy@biochem.uthscsa.edu<BR>
> ><BR>
> > University of Texas Health Science Center<BR>
> > Bioinformatics Core Facility<BR>
> > <A HREF="http://www.bioinformatics.uthscsa.edu">http://www.bioinformatics.uthscsa.edu</A><BR>
> > Phone: (210) 567-2672<BR>
> ><BR>
> ><BR>
> ><BR>
> > ------------------------------<BR>
> ><BR>
> > _______________________________________________<BR>
> > Lustre-discuss mailing list<BR>
> > Lustre-discuss@lists.lustre.org<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> ><BR>
> ><BR>
> > End of Lustre-discuss Digest, Vol 25, Issue 56<BR>
> > **********************************************<BR>
> ><BR>
> ><BR>
> > ------------------------------------------------------------------------<BR>
> ><BR>
> > _______________________________________________<BR>
> > Lustre-discuss mailing list<BR>
> > Lustre-discuss@lists.lustre.org<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
><BR>
><BR>
> --<BR>
> Amit Sharma<BR>
> Lustre Engineering<BR>
> Sun Microsystems, Bangalore.<BR>
><BR>
><BR>
> ------------------------------<BR>
><BR>
> Message: 3<BR>
> Date: Fri, 29 Feb 2008 12:27:53 +0530<BR>
> From: "ashok bharat bayana" <ashok.bharat.bayana@iiitb.ac.in><BR>
> Subject: Re: [Lustre-discuss] Configuration of lustre FS on single<BR>
> node<BR>
> To: "Amit Sharma" <Amit.Sharma@Sun.COM><BR>
> Cc: lustre-discuss@lists.lustre.org<BR>
> Message-ID: <8626C1B7EB748940BCDD7596134632BE85022A@jal.iiitb.ac.in><BR>
> Content-Type: text/plain; charset="iso-8859-1"<BR>
><BR>
><BR>
> Hi,<BR>
> the kernel version I used is 2.6.20.15 and the distro is Ubuntu 7.10<BR>
> The steps so far I have done are:<BR>
> 1)I have downloaded the sources lustre(1.6.4.2) from CFS<BR>
> 2)I tried to build lustre for a patchless client in which i triggered the command,<BR>
> [lustre]$ ./configure --with-linux=/unpatched/kernel/source<BR>
> 3)then I build it using make and then installed the modules using make install commands<BR>
> (I'm not sure whether the modules are installed correctly or not.Isn't the modules are installed using make install cmd)<BR>
> 4)Now in /usr/lib there is a lustre folder created.<BR>
> (Can we setup all MDS,client and OST's on a single system? my system configuration is Intel Pentium 512 MB RAM 1.6 GHz)<BR>
> Could you please tell me how to mount lustre fs in my laptop.<BR>
><BR>
> Thanks and Regards,<BR>
> Ashok Bharat<BR>
> -----Original Message-----<BR>
> From: Amit Sharma [<A HREF="mailto:Amit.Sharma@Sun.COM">mailto:Amit.Sharma@Sun.COM</A>]<BR>
> Sent: Fri 2/29/2008 12:03 PM<BR>
> To: ashok bharat bayana<BR>
> Cc: lustre-discuss@lists.lustre.org<BR>
> Subject: Re: [Lustre-discuss] Configuration of lustre FS on single node<BR>
> <BR>
> Ashok,<BR>
><BR>
> Could you let us know what are the modules that you have installed on<BR>
> the system. The linux version on your system. And the steps that you<BR>
> have done so far.<BR>
><BR>
> thanks,<BR>
> Amit<BR>
> ashok bharat bayana wrote:<BR>
> ><BR>
> > Hello,<BR>
> > I successfully build lustre(1.6.4.2) on my system for a patchless client<BR>
> > But I dont know how to proceed in configuring the file system.<BR>
> ><BR>
> > I'm trying to build all MDS,client and MDT's on a single node.<BR>
> > >From the tutorials I came to know that<BR>
> ><BR>
> > First, create an MDT for the "spfs" file system that uses the /dev/sda<BR>
> > disk. This MDT also acts as the MGS for the site.<BR>
> ><BR>
> > $ mkfs.lustre --fsname spfs --mdt --mgs /dev/sda<BR>
> ><BR>
> > But by triggering this command my system crashed in which the whole data<BR>
> > is lost and again I need to install the OS.<BR>
> > I want help in proceeding of mounting a lustre file system.<BR>
> ><BR>
> > Thanks and Regards,<BR>
> > Ashok Bharat<BR>
> ><BR>
> > -----Original Message-----<BR>
> > From: lustre-discuss-bounces@lists.lustre.org on behalf of<BR>
> > lustre-discuss-request@lists.lustre.org<BR>
> > Sent: Wed 2/27/2008 10:30 PM<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Subject: Lustre-discuss Digest, Vol 25, Issue 56<BR>
> ><BR>
> > Send Lustre-discuss mailing list submissions to<BR>
> > lustre-discuss@lists.lustre.org<BR>
> ><BR>
> > To subscribe or unsubscribe via the World Wide Web, visit<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> > or, via email, send a message with subject or body 'help' to<BR>
> > lustre-discuss-request@lists.lustre.org<BR>
> ><BR>
> > You can reach the person managing the list at<BR>
> > lustre-discuss-owner@lists.lustre.org<BR>
> ><BR>
> > When replying, please edit your Subject line so it is more specific<BR>
> > than "Re: Contents of Lustre-discuss digest..."<BR>
> ><BR>
> ><BR>
> > Today's Topics:<BR>
> ><BR>
> > 1. Multiple NICs per OST (Joshua Bower-Cooley)<BR>
> > 2. obd error on MGS/MDT node (Jeremy Mann)<BR>
> ><BR>
> ><BR>
> > ----------------------------------------------------------------------<BR>
> ><BR>
> > Message: 1<BR>
> > Date: Tue, 26 Feb 2008 11:55:19 -0800<BR>
> > From: Joshua Bower-Cooley <jbowercooley@lcogt.net><BR>
> > Subject: [Lustre-discuss] Multiple NICs per OST<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Message-ID: <200802261155.19907.jbowercooley@lcogt.net><BR>
> > Content-Type: text/plain; charset="utf-8"<BR>
> ><BR>
> > Hi-<BR>
> > I've had lustre running happily for some time over a single 10g ethernet NIC<BR>
> > per node. After switching to dual 10g and creating a new filesystem<BR>
> > (1.6.4.2), I'm seeing nothing but keep-alive packets with bad checksums.<BR>
> ><BR>
> > What is the current "correct" way to do this now? The manual suggest not<BR>
> > using<BR>
> > bonding, but several list postings now reccommend it. Without bonding, do I<BR>
> > need to have my 2 switches stacked, or will Lustre recognize the division in<BR>
> > my subnet?<BR>
> ><BR>
> > LNET module options I've tried are:<BR>
> > 1) networks="tcp0(eth2,eth3)"<BR>
> > 2) ip2nets="tcp(eth2,eth3); tcp(eth2) 10.9.[1-4].*; tcp(eth3) 10.9.[5-8].*;"<BR>
> > and many other variations<BR>
> ><BR>
> > All other services are functional on both interfaces.<BR>
> ><BR>
> > Thanks in advance,<BR>
> > J Bower-Cooley<BR>
> > Scientific Systems Engineer<BR>
> > Las Cumbres Observatory Global Telescope<BR>
> ><BR>
> ><BR>
> > ------------------------------<BR>
> ><BR>
> > Message: 2<BR>
> > Date: Wed, 27 Feb 2008 09:50:08 -0600 (CST)<BR>
> > From: "Jeremy Mann" <jeremy@biochem.uthscsa.edu><BR>
> > Subject: [Lustre-discuss] obd error on MGS/MDT node<BR>
> > To: lustre-discuss@lists.lustre.org<BR>
> > Message-ID:<BR>
> > <62901.24.173.62.147.1204127408.squirrel@biochem.uthscsa.edu><BR>
> > Content-Type: text/plain;charset=iso-8859-1<BR>
> ><BR>
> > Today we are starting to see obd_change_cbdata errors on our MGS/MDT node.<BR>
> > What does this error mean?<BR>
> ><BR>
> > Version of lustre and kernel is:<BR>
> ><BR>
> > 2.6.9-42.0.10.EL_lustre-1.6.0.1smp X86_64<BR>
> ><BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 2<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 4<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 5<BR>
> > previous similar messages<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata())<BR>
> > obd_change_cbdata: NULL export<BR>
> > LustreError: 65:0:(obd_class.h:1171:obd_change_cbdata()) Skipped 1<BR>
> > previous similar message<BR>
> ><BR>
> > --<BR>
> > Jeremy Mann<BR>
> > jeremy@biochem.uthscsa.edu<BR>
> ><BR>
> > University of Texas Health Science Center<BR>
> > Bioinformatics Core Facility<BR>
> > <A HREF="http://www.bioinformatics.uthscsa.edu">http://www.bioinformatics.uthscsa.edu</A><BR>
> > Phone: (210) 567-2672<BR>
> ><BR>
> ><BR>
> ><BR>
> > ------------------------------<BR>
> ><BR>
> > _______________________________________________<BR>
> > Lustre-discuss mailing list<BR>
> > Lustre-discuss@lists.lustre.org<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
> ><BR>
> ><BR>
> > End of Lustre-discuss Digest, Vol 25, Issue 56<BR>
> > **********************************************<BR>
> ><BR>
> ><BR>
> > ------------------------------------------------------------------------<BR>
> ><BR>
> > _______________________________________________<BR>
> > Lustre-discuss mailing list<BR>
> > Lustre-discuss@lists.lustre.org<BR>
> > <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
><BR>
><BR>
> --<BR>
> Amit Sharma<BR>
> Lustre Engineering<BR>
> Sun Microsystems, Bangalore.<BR>
><BR>
> -------------- next part --------------<BR>
> An HTML attachment was scrubbed...<BR>
> URL: <A HREF="http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080229/39e94cf4/attachment.html">http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080229/39e94cf4/attachment.html</A><BR>
><BR>
> ------------------------------<BR>
><BR>
> _______________________________________________<BR>
> Lustre-discuss mailing list<BR>
> Lustre-discuss@lists.lustre.org<BR>
> <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
><BR>
><BR>
> End of Lustre-discuss Digest, Vol 25, Issue 58<BR>
> **********************************************<BR>
><BR>
<BR>
> _______________________________________________<BR>
> Lustre-discuss mailing list<BR>
> Lustre-discuss@lists.lustre.org<BR>
> <A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
<BR>
<BR>
Cheers, Andreas<BR>
--<BR>
Andreas Dilger<BR>
Sr. Staff Engineer, Lustre Group<BR>
Sun Microsystems of Canada, Inc.<BR>
<BR>
<BR>
<BR>
------------------------------<BR>
<BR>
Message: 2<BR>
Date: Mon, 10 Mar 2008 11:08:18 +0200<BR>
From: Alex Lyashkov <Alexey.Lyashkov@Sun.COM><BR>
Subject: Re: [Lustre-discuss] yet another lustre error<BR>
To: Brock Palen <brockp@umich.edu><BR>
Cc: lustre-discuss@lists.lustre.org<BR>
Message-ID: <1205140098.1359.270.camel@bear.shadowland><BR>
Content-Type: text/plain<BR>
<BR>
On Fri, 2008-03-07 at 18:45 -0500, Brock Palen wrote:<BR>
> On a file system thats been up for only 57 days, I have:<BR>
><BR>
<BR>
> target_handle_reconnect<BR>
> nobackup-MDT0000: 34b4fbea-200b-1f7c-dac0-516b8ce786fc reconnecting<BR>
> ldlm_lib.c<BR>
> target_handle_connect<BR>
> nobackup-MDT0000: refuse reconnection from 34b4fbea-200b-1f7c-<BR>
> dac0-516b8ce786fc@10.164.0.111@tcp to 0x00000100069a7000; still busy <BR>
> with 2 active RPCs<BR>
> ldlm_lib.c<BR>
> target_send_reply_msg<BR>
> @@@ processing error (-16) req@0000010019159a00 x11199816/t0 o38-<BR>
> >34b4fbea-200b-1f7c-dac0-516b8ce786fc@NET_0x200000aa4006f_UUID:-1 <BR>
> lens 304/200 ref 0 fl Interpret:/0/0 rc -16/0<BR>
><BR>
><BR>
> What I see messages about active rpc's in other logs. What would <BR>
> this mean? Is something suck someplace ?<BR>
><BR>
-16 = EBUSY. This say client reconnected to server which already work on<BR>
different request from this client. After old rpc from this client will<BR>
be finished - client will be reconnected.<BR>
<BR>
--<BR>
Alex Lyashkov <Alexey.lyashkov@sun.com><BR>
Lustre Group, Sun Microsystems<BR>
<BR>
<BR>
<BR>
------------------------------<BR>
<BR>
Message: 3<BR>
Date: Mon, 10 Mar 2008 12:27:51 +0100<BR>
From: Joe Barjo <jobarjo78@yahoo.fr><BR>
Subject: Re: [Lustre-discuss] lustre and small files overhead<BR>
Cc: lustre-discuss@lists.lustre.org<BR>
Message-ID: <47D51B37.5070703@yahoo.fr><BR>
Content-Type: text/plain; charset="iso-8859-1"<BR>
<BR>
Andreas Dilger a ?crit :<BR>
> On Mar 07, 2008 12:49 +0100, Joe Barjo wrote:<BR>
> <BR>
>> I made some more tests, and have setup a micro lustre cluster on lvm<BR>
>> volumes.<BR>
>> node a: MDS<BR>
>> node b and c: OST<BR>
>> node a,b,c,d,e,f: clients<BR>
>> Gigabit ethernet network.<BR>
>> Made the optimizations: lnet.debug=0, lru_size to 10000, max_dirty_mb to<BR>
>> 1024<BR>
>> <BR>
><BR>
> For high RPC-rate operations using an interconnect like Infiniband is<BR>
> better than ethernet.<BR>
><BR>
> <BR>
infiniband is not in our budget...<BR>
>> The svn checkout takes 50s ( 15s on a localdisk, 25s on a local lustre<BR>
>> demo (with debug=0))<BR>
>> Launching gkrellm, a single svn checkout consumes about 20% of the MDS<BR>
>> system cpu with about 2.4mbyte/sec ethernet communication.<BR>
>> <BR>
><BR>
> <BR>
>> About 6MByte/s disk bandwidth on OST1, up to 12-16MB/s on OST2 disk<BR>
>> bandwidth, network bandwidth on OST is about 10 to 20 times under disk<BR>
>> bandwidth.<BR>
>> Why so much disk bandwidth on OSTs, is it a readahead problem?<BR>
>> <BR>
><BR>
> That does seem strange, I can't really say why. There is some metadata<BR>
> overhead, and that is higher with small files, but I don't think it<BR>
> would be 10-20x overhead.<BR>
><BR>
> <BR>
The checkouted source is only 65 megabytes. So much OST disk bandwidth<BR>
is probably not normal.<BR>
Maybe you should verify this point.<BR>
Are you sure there isn't an optimazation for this? This looks like<BR>
readahead or something like that.<BR>
>> I launched a compilation distributed on the 6 clients:<BR>
>> MDS system cpu goes up to 60% system ressource (athlon 64 3500+)<BR>
>> 12MByte/s on the ethernet, OST goes up to the same level as previous test.<BR>
>><BR>
>> How come is there so much network communications on the MDT?<BR>
>> <BR>
><BR>
> Because every metadata operation has to be done on the MDS currently.<BR>
> We are working toward having metadata writeback cache operations on<BR>
> the client, but it doesn't happen currently. For operations like<BR>
> compilation it is basically entirely metadata overhead.<BR>
><BR>
> <BR>
>> As I understood that the MDS can not be load balanced, I don't see how<BR>
>> lustre is scalable to thousands of clients...<BR>
>> <BR>
><BR>
> Because in many HPC environments there are very few metadata operations<BR>
> in comparison to the amount of data being read/written. Average file<BR>
> sizes are 20-30MB instead of 20-30kB.<BR>
><BR>
> <BR>
>> It looks like lustre is not made for this kind of application<BR>
>> <BR>
><BR>
> No, it definitely isn't tuned for small files.<BR>
> <BR>
Could it be tuned one day for small files?<BR>
Which filesystem would you suggest for me?<BR>
I already tried nfs, afs<BR>
I will now try glusterfs<BR>
<BR>
Thanks for your support<BR>
<BR>
-------------- next part --------------<BR>
An HTML attachment was scrubbed...<BR>
URL: <A HREF="http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080310/11e3dfa0/attachment.html">http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080310/11e3dfa0/attachment.html</A><BR>
<BR>
------------------------------<BR>
<BR>
_______________________________________________<BR>
Lustre-discuss mailing list<BR>
Lustre-discuss@lists.lustre.org<BR>
<A HREF="http://lists.lustre.org/mailman/listinfo/lustre-discuss">http://lists.lustre.org/mailman/listinfo/lustre-discuss</A><BR>
<BR>
<BR>
End of Lustre-discuss Digest, Vol 26, Issue 25<BR>
**********************************************<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>