Hi, I''m new to Lustre and trying to evaluate 1.6 beta. My first goal is to setup a single node system. I''m following the MountConf manual and a bit confused by the section "Module setup" (not sure where to find utils/module_setup, and which "make all" it''s talking about). Do I really need to do anything here? Or can I just run mkfs.lustre and mount to get the system up and running? Thanks lin -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20061115/1f8f5774/attachment.html
The script lustre/tests/llmount.sh will load the modules from the local build dir and start a small Lustre fs on a single node. You can change many of options by overriding the values set in lustre/tests/cfg/local.sh if e.g. you want to use real disks. But eventually you need to put the modules in the appropriate place for the kernel to find them if you ever want to do anything else. lustre/utils/module_setup.sh does this for you, and modifies /etc/modprobe.conf for a basic tcp LNET setup. Lin Shen (lshen) wrote:> > Hi, > > I''m new to Lustre and trying to evaluate 1.6 beta. My first goal is to > setup a single node system. I''m following the MountConf manual and a > bit confused by the section "Module setup" (not sure where to find > utils/module_setup, and which "make all" it''s talking about). Do I > really need to do anything here? Or can I just run mkfs.lustre and > mount to get the system up and running? > > Thanks > lin > > ------------------------------------------------------------------------ > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >
llmount.sh seems to be missing pieces (cfg/local.sh for example) in 1.5.95. Is there a newer 1.6 beta release where it works? Also, I don''t think the MountConf document mentions that you need to replace your e2fsprogs with the CFS-modified version or you''ll be chasing cryptic errors. Jim On Wed, 15 Nov 2006, Nathaniel Rutman wrote:> The script lustre/tests/llmount.sh will load the modules from the local build > dir and start a small Lustre fs on a single node. You can change many of > options by overriding the values set in lustre/tests/cfg/local.sh if e.g. you > want to use real disks. > But eventually you need to put the modules in the appropriate place for the > kernel to find them if you ever want to do anything else. > lustre/utils/module_setup.sh does this for you, and modifies > /etc/modprobe.conf for a basic tcp LNET setup. > > > Lin Shen (lshen) wrote: >> >> Hi, >> >> I''m new to Lustre and trying to evaluate 1.6 beta. My first goal is to >> setup a single node system. I''m following the MountConf manual and a bit >> confused by the section "Module setup" (not sure where to find >> utils/module_setup, and which "make all" it''s talking about). Do I really >> need to do anything here? Or can I just run mkfs.lustre and mount to get >> the system up and running? >> >> Thanks >> lin >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Lustre-discuss mailing list >> Lustre-discuss@clusterfs.com >> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >> > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >
I didn''t build Lustre. I''m installing the pre-packaged RPMs. Do I still need to run those scripts? And I don''t think I even have them since I didn''t install the *source*> -----Original Message----- > From: Nathaniel Rutman [mailto:nathan@clusterfs.com] > Sent: Wednesday, November 15, 2006 12:26 PM > To: Lin Shen (lshen) > Cc: lustre-discuss@clusterfs.com > Subject: Re: [Lustre-discuss] Configure 1.6 beta > > The script lustre/tests/llmount.sh will load the modules from > the local build dir and start a small Lustre fs on a single > node. You can change many of options by overriding the > values set in lustre/tests/cfg/local.sh if e.g. you want to > use real disks. > But eventually you need to put the modules in the appropriate > place for the kernel to find them if you ever want to do > anything else. > lustre/utils/module_setup.sh does this for you, and modifies > /etc/modprobe.conf for a basic tcp LNET setup. > > > Lin Shen (lshen) wrote: > > > > Hi, > > > > I''m new to Lustre and trying to evaluate 1.6 beta. My first > goal is to > > setup a single node system. I''m following the MountConf > manual and a > > bit confused by the section "Module setup" (not sure where to find > > utils/module_setup, and which "make all" it''s talking about). Do I > > really need to do anything here? Or can I just run mkfs.lustre and > > mount to get the system up and running? > > > > Thanks > > lin > > > > > ---------------------------------------------------------------------- > > -- > > > > _______________________________________________ > > Lustre-discuss mailing list > > Lustre-discuss@clusterfs.com > > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss > > >
Another related question. What do I need to do if I want to roll back to 1.4.7.3 from 1.6? lin> -----Original Message----- > From: Lin Shen (lshen) > Sent: Wednesday, November 15, 2006 1:18 PM > To: ''Nathaniel Rutman'' > Cc: lustre-discuss@clusterfs.com > Subject: RE: [Lustre-discuss] Configure 1.6 beta > > I didn''t build Lustre. I''m installing the pre-packaged RPMs. > Do I still need to run those scripts? And I don''t think I > even have them since I didn''t install the *source* > > > -----Original Message----- > > From: Nathaniel Rutman [mailto:nathan@clusterfs.com] > > Sent: Wednesday, November 15, 2006 12:26 PM > > To: Lin Shen (lshen) > > Cc: lustre-discuss@clusterfs.com > > Subject: Re: [Lustre-discuss] Configure 1.6 beta > > > > The script lustre/tests/llmount.sh will load the modules from the > > local build dir and start a small Lustre fs on a single > node. You can > > change many of options by overriding the values set in > > lustre/tests/cfg/local.sh if e.g. you want to use real disks. > > But eventually you need to put the modules in the appropriate place > > for the kernel to find them if you ever want to do anything else. > > lustre/utils/module_setup.sh does this for you, and modifies > > /etc/modprobe.conf for a basic tcp LNET setup. > > > > > > Lin Shen (lshen) wrote: > > > > > > Hi, > > > > > > I''m new to Lustre and trying to evaluate 1.6 beta. My first > > goal is to > > > setup a single node system. I''m following the MountConf > > manual and a > > > bit confused by the section "Module setup" (not sure > where to find > > > utils/module_setup, and which "make all" it''s talking > about). Do I > > > really need to do anything here? Or can I just run > mkfs.lustre and > > > mount to get the system up and running? > > > > > > Thanks > > > lin > > > > > > > > > ---------------------------------------------------------------------- > > > -- > > > > > > _______________________________________________ > > > Lustre-discuss mailing list > > > Lustre-discuss@clusterfs.com > > > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss > > > > > >
Jim Garlick wrote:> llmount.sh seems to be missing pieces (cfg/local.sh for example) in > 1.5.95. Is there a newer 1.6 beta release where it works? >Wow, you''re right, apparently the cfg dir is not put in the distribution tarball. I''ll fix that.> Also, I don''t think the MountConf document mentions that you need to > replace your e2fsprogs with the CFS-modified version or you''ll be > chasing cryptic errors. >Thank you, docs fixed. Hopefully the errors aren''t too cryptic - I tried to say "hey, use our e2fsprogs" where those errors occur. If there are places I missed, feel free to point them out to me.
Lin Shen (lshen) wrote:> I didn''t build Lustre. I''m installing the pre-packaged RPMs. Do I still > need to run those scripts? And I don''t think I even have them since I > didn''t install the *source* > >llmount is only in the tarball, not the rpms; follow the instructions on the wiki for a simple setup. The rpms install the modules in the right place. modprobe.conf only needs to be modified to work with your networking setup, see https://mail.clusterfs.com/wikis/lustre/MountConf#head-84a5d70d2684c8bfcbc203119a56b44cb2421e2b> Another related question. What do I need to do if I want to roll back to > 1.4.7.3 from 1.6? > >There''s a section on downgrading at https://mail.clusterfs.com/wikis/lustre/UpgradeTo16>> -----Original Message----- >> From: Nathaniel Rutman [mailto:nathan@clusterfs.com] >> Sent: Wednesday, November 15, 2006 12:26 PM >> To: Lin Shen (lshen) >> Cc: lustre-discuss@clusterfs.com >> Subject: Re: [Lustre-discuss] Configure 1.6 beta >> >> The script lustre/tests/llmount.sh will load the modules from >> the local build dir and start a small Lustre fs on a single >> node. You can change many of options by overriding the >> values set in lustre/tests/cfg/local.sh if e.g. you want to >> use real disks. >> But eventually you need to put the modules in the appropriate >> place for the kernel to find them if you ever want to do >> anything else. >> lustre/utils/module_setup.sh does this for you, and modifies >> /etc/modprobe.conf for a basic tcp LNET setup. >> >> >> Lin Shen (lshen) wrote: >> >>> Hi, >>> >>> I''m new to Lustre and trying to evaluate 1.6 beta. My first >>> >> goal is to >> >>> setup a single node system. I''m following the MountConf >>> >> manual and a >> >>> bit confused by the section "Module setup" (not sure where to find >>> utils/module_setup, and which "make all" it''s talking about). Do I >>> really need to do anything here? Or can I just run mkfs.lustre and >>> mount to get the system up and running? >>> >>> Thanks >>> lin >>> >>> >>> >> ---------------------------------------------------------------------- >> >>> -- >>> >>> _______________________________________________ >>> Lustre-discuss mailing list >>> Lustre-discuss@clusterfs.com >>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>> >>> > >
Hi Nathan, Have we considered making the module loading an option to mount.lustre - e.g. mount -t lustre -o moddir=/foo/bar .... To make the shell script almost go away. I think this would be very nice. - Peter -> -----Original Message----- > From: lustre-discuss-bounces@clusterfs.com > [mailto:lustre-discuss-bounces@clusterfs.com] On Behalf Of > Nathaniel Rutman > Sent: Wednesday, November 15, 2006 1:26 PM > To: Lin Shen (lshen) > Cc: lustre-discuss@clusterfs.com > Subject: Re: [Lustre-discuss] Configure 1.6 beta > > The script lustre/tests/llmount.sh will load the modules from > the local build dir and start a small Lustre fs on a single > node. You can change many of options by overriding the > values set in lustre/tests/cfg/local.sh if e.g. you want to > use real disks. > But eventually you need to put the modules in the appropriate > place for the kernel to find them if you ever want to do > anything else. > lustre/utils/module_setup.sh does this for you, and modifies > /etc/modprobe.conf for a basic tcp LNET setup. > > > Lin Shen (lshen) wrote: > > > > Hi, > > > > I''m new to Lustre and trying to evaluate 1.6 beta. My first > goal is to > > setup a single node system. I''m following the MountConf > manual and a > > bit confused by the section "Module setup" (not sure where to find > > utils/module_setup, and which "make all" it''s talking about). Do I > > really need to do anything here? Or can I just run mkfs.lustre and > > mount to get the system up and running? > > > > Thanks > > lin > > > > > ---------------------------------------------------------------------- > > -- > > > > _______________________________________________ > > Lustre-discuss mailing list > > Lustre-discuss@clusterfs.com > > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss > > > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
On Wed, 15 Nov 2006, Nathaniel Rutman wrote:> Jim Garlick wrote: >> Also, I don''t think the MountConf document mentions that you need to >> replace your e2fsprogs with the CFS-modified version or you''ll be >> chasing cryptic errors. >> > Thank you, docs fixed. Hopefully the errors aren''t too cryptic - I tried to > say "hey, use our e2fsprogs" where those errors occur. If there are places I > missed, feel free to point them out to me.A suggestion: you could remove this stumbling block by including a copy of the modified e2fsprogs as ''ldiskfsprogs'' in the lustre source, packaged so they don''t conflict with the distro''s e2fsprogs of course... Jim
The appropriate modules are loaded automatically, on demand, when "mount -t lustre" is given. The only reason llmount (and more generally the test-framework.sh) tries to load from a local dir is because certain CFS developers don''t like to copy their in-development modules into the place where they are _supposed_ to live (/lib/modules/`uname -r`/). No script is needed in the "normal" case (rpms or ''make install'') Peter J. Braam wrote:> Hi Nathan, > > Have we considered making the module loading an option to mount.lustre - > e.g. > > mount -t lustre -o moddir=/foo/bar .... > > To make the shell script almost go away. I think this would be very nice. > > - Peter - > >>> -----Original Message----- >> From: lustre-discuss-bounces@clusterfs.com >> [mailto:lustre-discuss-bounces@clusterfs.com] On Behalf Of >> Nathaniel Rutman >> Sent: Wednesday, November 15, 2006 1:26 PM >> To: Lin Shen (lshen) >> Cc: lustre-discuss@clusterfs.com >> Subject: Re: [Lustre-discuss] Configure 1.6 beta >> >> The script lustre/tests/llmount.sh will load the modules from >> the local build dir and start a small Lustre fs on a single >> node. You can change many of options by overriding the >> values set in lustre/tests/cfg/local.sh if e.g. you want to >> use real disks. >> But eventually you need to put the modules in the appropriate >> place for the kernel to find them if you ever want to do >> anything else. >> lustre/utils/module_setup.sh does this for you, and modifies >> /etc/modprobe.conf for a basic tcp LNET setup. >> >> >> Lin Shen (lshen) wrote: >> >>> Hi, >>> >>> I''m new to Lustre and trying to evaluate 1.6 beta. My first >>> >> goal is to >> >>> setup a single node system. I''m following the MountConf >>> >> manual and a >> >>> bit confused by the section "Module setup" (not sure where to find >>> utils/module_setup, and which "make all" it''s talking about). Do I >>> really need to do anything here? Or can I just run mkfs.lustre and >>> mount to get the system up and running? >>> >>> Thanks >>> lin >>> >>> >>> >> ---------------------------------------------------------------------- >> >>> -- >>> >>> _______________________________________________ >>> Lustre-discuss mailing list >>> Lustre-discuss@clusterfs.com >>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>> >>> >> _______________________________________________ >> Lustre-discuss mailing list >> Lustre-discuss@clusterfs.com >> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >> > >
One of my tasks for LLNL is to get the ''sanity'' tests to work against packaged and installed modules in our QA environment. However, I think being able to run the tests on modules located in the source tree has merit too as building an rpm implies a rebuild from scratch, and that is a pain when you are making small changes and testing as you go. (running ''make install'' is strictly verboten here as it is too easy to violate someone elses packaging and screw up something - like QA results or production!) So, my $0.02 (for what its worth) is I kinda like Peter''s suggestion. Jim On Wed, 15 Nov 2006, Nathaniel Rutman wrote:> > The appropriate modules are loaded automatically, on demand, when "mount -t > lustre" is given. The only reason llmount (and more generally the > test-framework.sh) tries to load from a local dir is because certain CFS > developers don''t like to copy their in-development modules into the place > where they are _supposed_ to live (/lib/modules/`uname -r`/). No script is > needed in the "normal" case (rpms or ''make install'') > > > Peter J. Braam wrote: >> Hi Nathan, >> >> Have we considered making the module loading an option to mount.lustre - >> e.g. >> mount -t lustre -o moddir=/foo/bar .... To make the shell script almost go >> away. I think this would be very nice. >> >> - Peter - >> >> > >>> -----Original Message----- >>> From: lustre-discuss-bounces@clusterfs.com >>> [mailto:lustre-discuss-bounces@clusterfs.com] On Behalf Of Nathaniel >>> Rutman >>> Sent: Wednesday, November 15, 2006 1:26 PM >>> To: Lin Shen (lshen) >>> Cc: lustre-discuss@clusterfs.com >>> Subject: Re: [Lustre-discuss] Configure 1.6 beta >>> >>> The script lustre/tests/llmount.sh will load the modules from the local >>> build dir and start a small Lustre fs on a single node. You can change >>> many of options by overriding the values set in lustre/tests/cfg/local.sh >>> if e.g. you want to use real disks. >>> But eventually you need to put the modules in the appropriate place for >>> the kernel to find them if you ever want to do anything else. >>> lustre/utils/module_setup.sh does this for you, and modifies >>> /etc/modprobe.conf for a basic tcp LNET setup. >>> >>> >>> Lin Shen (lshen) wrote: >>> >>>> Hi, >>>> >>>> I''m new to Lustre and trying to evaluate 1.6 beta. My first >>> goal is to >>>> setup a single node system. I''m following the MountConf >>> manual and a >>>> bit confused by the section "Module setup" (not sure where to find >>>> utils/module_setup, and which "make all" it''s talking about). Do I really >>>> need to do anything here? Or can I just run mkfs.lustre and mount to get >>>> the system up and running? >>>> >>>> Thanks >>>> lin >>>> >>>> >>>> >>> ---------------------------------------------------------------------- >>> >>>> -- >>>> >>>> _______________________________________________ >>>> Lustre-discuss mailing list >>>> Lustre-discuss@clusterfs.com >>>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>>> >>> _______________________________________________ >>> Lustre-discuss mailing list >>> Lustre-discuss@clusterfs.com >>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>> >> >> > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >
You can still use the minimal-impact lustre/utils/module_setup.sh script without having to make install. Additionally, lustre/tests/sanity.sh also uses test-framework, and so will insmod modules from the local directory (unless env. variable MODPROBE=true) anyhow. Further problems with not using the correct kernel modules directory is that you must use insmod instead of modprobe, which 1. requires a specific module loading order, and 2. will ignore any settings in modprobe.conf, e.g. options lnet networks=(anything other than tcp eth0) Jim Garlick wrote:> One of my tasks for LLNL is to get the ''sanity'' tests to work against > packaged and installed modules in our QA environment. > > However, I think being able to run the tests on modules located in the > source tree has merit too as building an rpm implies a rebuild from > scratch, and that is a pain when you are making small changes and > testing as you go. (running ''make install'' is strictly > verboten here as it is too easy to violate someone elses packaging and > screw up something - like QA results or production!) > > So, my $0.02 (for what its worth) is I kinda like Peter''s suggestion. > > Jim > > On Wed, 15 Nov 2006, Nathaniel Rutman wrote: > >> >> The appropriate modules are loaded automatically, on demand, when >> "mount -t lustre" is given. The only reason llmount (and more >> generally the test-framework.sh) tries to load from a local dir is >> because certain CFS developers don''t like to copy their >> in-development modules into the place where they are _supposed_ to >> live (/lib/modules/`uname -r`/). No script is needed in the "normal" >> case (rpms or ''make install'') >> >> >> Peter J. Braam wrote: >>> Hi Nathan, >>> >>> Have we considered making the module loading an option to >>> mount.lustre - >>> e.g. mount -t lustre -o moddir=/foo/bar .... To make the shell >>> script almost go away. I think this would be very nice. >>> >>> - Peter - >>> >>> >> >>>> -----Original Message----- >>>> From: lustre-discuss-bounces@clusterfs.com >>>> [mailto:lustre-discuss-bounces@clusterfs.com] On Behalf Of >>>> Nathaniel Rutman >>>> Sent: Wednesday, November 15, 2006 1:26 PM >>>> To: Lin Shen (lshen) >>>> Cc: lustre-discuss@clusterfs.com >>>> Subject: Re: [Lustre-discuss] Configure 1.6 beta >>>> >>>> The script lustre/tests/llmount.sh will load the modules from the >>>> local build dir and start a small Lustre fs on a single node. You >>>> can change many of options by overriding the values set in >>>> lustre/tests/cfg/local.sh if e.g. you want to use real disks. >>>> But eventually you need to put the modules in the appropriate place >>>> for the kernel to find them if you ever want to do anything else. >>>> lustre/utils/module_setup.sh does this for you, and modifies >>>> /etc/modprobe.conf for a basic tcp LNET setup. >>>> >>>> >>>> Lin Shen (lshen) wrote: >>>> >>>>> Hi, >>>>> >>>>> I''m new to Lustre and trying to evaluate 1.6 beta. My first >>>> goal is to >>>>> setup a single node system. I''m following the MountConf >>>> manual and a >>>>> bit confused by the section "Module setup" (not sure where to find >>>>> utils/module_setup, and which "make all" it''s talking about). Do I >>>>> really need to do anything here? Or can I just run mkfs.lustre and >>>>> mount to get the system up and running? >>>>> >>>>> Thanks >>>>> lin >>>>> >>>>> >>>>> >>>> ---------------------------------------------------------------------- >>>> >>>>> -- >>>>> >>>>> _______________________________________________ >>>>> Lustre-discuss mailing list >>>>> Lustre-discuss@clusterfs.com >>>>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>>>> >>>> _______________________________________________ >>>> Lustre-discuss mailing list >>>> Lustre-discuss@clusterfs.com >>>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >>>> >>> >>> >> >> _______________________________________________ >> Lustre-discuss mailing list >> Lustre-discuss@clusterfs.com >> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >> >
Nathan, Yes, I understand and it''s precisely the developer & test case that I''d like to cover without a wrapper. - Peter -> -----Original Message----- > From: Nathaniel Rutman [mailto:nathan@clusterfs.com] > Sent: Wednesday, November 15, 2006 6:41 PM > To: Peter J. Braam > Cc: ''Lin Shen (lshen)''; lustre-discuss@clusterfs.com > Subject: Re: [Lustre-discuss] Configure 1.6 beta > > > The appropriate modules are loaded automatically, on demand, > when "mount -t lustre" is given. The only reason llmount > (and more generally the > test-framework.sh) tries to load from a local dir is because > certain CFS developers don''t like to copy their > in-development modules into the place where they are > _supposed_ to live (/lib/modules/`uname -r`/). > No script is needed in the "normal" case (rpms or ''make install'') > > > Peter J. Braam wrote: > > Hi Nathan, > > > > Have we considered making the module loading an option to > mount.lustre > > - e.g. > > > > mount -t lustre -o moddir=/foo/bar .... > > > > To make the shell script almost go away. I think this > would be very nice. > > > > - Peter - > > > > > > >> -----Original Message----- > >> From: lustre-discuss-bounces@clusterfs.com > >> [mailto:lustre-discuss-bounces@clusterfs.com] On Behalf Of > Nathaniel > >> Rutman > >> Sent: Wednesday, November 15, 2006 1:26 PM > >> To: Lin Shen (lshen) > >> Cc: lustre-discuss@clusterfs.com > >> Subject: Re: [Lustre-discuss] Configure 1.6 beta > >> > >> The script lustre/tests/llmount.sh will load the modules from the > >> local build dir and start a small Lustre fs on a single node. You > >> can change many of options by overriding the values set in > >> lustre/tests/cfg/local.sh if e.g. you want to use real disks. > >> But eventually you need to put the modules in the > appropriate place > >> for the kernel to find them if you ever want to do anything else. > >> lustre/utils/module_setup.sh does this for you, and modifies > >> /etc/modprobe.conf for a basic tcp LNET setup. > >> > >> > >> Lin Shen (lshen) wrote: > >> > >>> Hi, > >>> > >>> I''m new to Lustre and trying to evaluate 1.6 beta. My first > >>> > >> goal is to > >> > >>> setup a single node system. I''m following the MountConf > >>> > >> manual and a > >> > >>> bit confused by the section "Module setup" (not sure > where to find > >>> utils/module_setup, and which "make all" it''s talking > about). Do I > >>> really need to do anything here? Or can I just run > mkfs.lustre and > >>> mount to get the system up and running? > >>> > >>> Thanks > >>> lin > >>> > >>> > >>> > >> > --------------------------------------------------------------------- > >> - > >> > >>> -- > >>> > >>> _______________________________________________ > >>> Lustre-discuss mailing list > >>> Lustre-discuss@clusterfs.com > >>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss > >>> > >>> > >> _______________________________________________ > >> Lustre-discuss mailing list > >> Lustre-discuss@clusterfs.com > >> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss > >> > > > >
On Wed, 15 Nov 2006, Nathaniel Rutman wrote:> Jim Garlick wrote: >> llmount.sh seems to be missing pieces (cfg/local.sh for example) in >> 1.5.95. Is there a newer 1.6 beta release where it works? >> > Wow, you''re right, apparently the cfg dir is not put in the distribution > tarball. I''ll fix that.Hi Nathan, Where can I pick up the missing bits? Thanks, Jim
Attaching cfg/local.sh here. Jim Garlick wrote:> On Wed, 15 Nov 2006, Nathaniel Rutman wrote: > >> Jim Garlick wrote: >>> llmount.sh seems to be missing pieces (cfg/local.sh for example) in >>> 1.5.95. Is there a newer 1.6 beta release where it works? >>> >> Wow, you''re right, apparently the cfg dir is not put in the >> distribution tarball. I''ll fix that. > > Hi Nathan, > > Where can I pick up the missing bits? > > Thanks, > > Jim >-------------- next part -------------- A non-text attachment was scrubbed... Name: local.sh Type: application/x-shellscript Size: 2287 bytes Desc: not available Url : http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20061120/7f53806d/local.bin