Felix, Evan J wrote:
>Peter, what do you see as the purpose of this list, is it the master
>Lustre development list like the Linux kernel development list, or will
>it be a ''external'' Lustre development list like its
predecessor. One of
>my dislikes about the old list, was that it was not the list y''all
used
>for your internal discussions of features, problems, etc. Are we to
>post patches and bugs here, or do we use bugzilla still? Also how will
>this list be different than the discuss list?
>
>and since we just got the 1.6 list I have a few questions/bugs:
>
>1. The old lconf would detect I was running out of a development tree,
>and load the modules for me? what is the new procedure. mkfs.Lustre
>just complains the modules are not present. Or is there some script
I''m
>missing?
>
>
>
The modules have to be in the appropriate /lib/modules directory for the
running kernel. You can run the utils/module_setup.sh script to copy
them over easily.
Then, as long as your modprobe.conf is set up, all modules should be
loaded automatically, when needed.
Sample modprobe.conf:
# Networking options, see /sys/module/lnet/parameters
options lnet networks=tcp
alias lustre llite
# end Lustre modules
mkfs.lustre doesn''t use any modules except ldiskfs, which should be
loaded automatically.
mount.lustre will load any required modules as well:
uml1:/r/home/nathan/cfs/b1_4_mountconf/lustre/utils# lsmod
Module Size Used by
uml1:/r/home/nathan/cfs/b1_4_mountconf/lustre/utils# mount -t lustre -o
loop /r/tmp/mdt /mnt/mdt
uml1:/r/home/nathan/cfs/b1_4_mountconf/lustre/utils# lsmod
Module Size Used by
mds 1382040 2
fsfilt_ldiskfs 135476 2
mgs 441120 1
mgc 138596 2 mgs
ldiskfs 302560 2 fsfilt_ldiskfs
llite 1168932 0
lov 782240 2 llite
mdc 292720 1 llite
ksocklnd 203872 1
ptlrpc 2102448 6 mds,mgs,mgc,llite,lov,mdc
obdclass 2726936 9
mds,fsfilt_ldiskfs,mgs,mgc,llite,lov,mdc,ptlrpc
lnet 478480 3 ksocklnd,ptlrpc,obdclass
lvfs 124584 10
mds,fsfilt_ldiskfs,mgs,mgc,llite,lov,mdc,ptlrpc,obdclass
libcfs 191296 12
mds,fsfilt_ldiskfs,mgs,mgc,llite,lov,mdc,ksocklnd,ptlrpc,obdclass,lnet,lvfs
>2. It seems that mkfs.lustre does not like using full devices (ie
>/dev/sda) as it does not pass the -F flag to mkfs.ext2. Can you enable
>this functionality, or are you expecting people will use
>--mkfsoptions="-F" when the make filesystems on full devices?
>
>
>
I''ll fix that:
RCS file: /cvsroot/cfs/lustre-core/utils/Attic/mkfs_lustre.c,v
retrieving revision 1.1.12.17.2.44
diff -u -p -r1.1.12.17.2.44 mkfs_lustre.c
--- mkfs_lustre.c 27 Mar 2006 23:04:40 -0000 1.1.12.17.2.44
+++ mkfs_lustre.c 19 Apr 2006 18:45:59 -0000
@@ -461,6 +461,11 @@ int make_lustre_backfs(struct mkfs_opts
strcat(mop->mo_mkfsopts, " -O dir_index");
}
+ /* Allow reformat of full devices */
+ if (!(mop->mo_flags & MO_IS_LOOP)) {
+ strcat(mop->mo_mkfsopts, " -F");
+ }
+
sprintf(mkfs_cmd, "mkfs.ext2 -j -b %d -L %s ",
L_BLOCK_SIZE,
mop->mo_ldd.ldd_svname);
>I have been able to mount a 1.6 system under Xen using 2.6.12.5 and it
>seems to work so far.
>
>
Good, I hope you like what you see.
>Evan
>________________________________
>
>From: lustre-devel-bounces@clusterfs.com
>[mailto:lustre-devel-bounces@clusterfs.com] On Behalf Of peter braam
>Sent: Wednesday, April 19, 2006 6:58 AM
>To: lustre-devel@clusterfs.com
>Subject: [Lustre-devel] welcome
>
>
>Hello
>
>Welcome to the Lustre devel mailing list, which we resurrected to
>discuss development efforts with the community. The lively discussion
>on lustre-discuss makes us believe this could bear fruits in the near
>future.
>
>- Peter -
>_______________________________________________
>Lustre-devel mailing list
>Lustre-devel@clusterfs.com
>https://mail.clusterfs.com/mailman/listinfo/lustre-devel
>
>