Whamcloud - gitweb
fs/lustre-release.git
18 years agoReduce duplication in and simplify.
brian [Fri, 30 Jun 2006 16:11:30 +0000 (16:11 +0000)]
Reduce duplication in and simplify.

18 years agoMerge b1_5 from b1_4 (20060630_1438)
vitaly [Fri, 30 Jun 2006 15:05:36 +0000 (15:05 +0000)]
Merge b1_5 from b1_4 (20060630_1438)

18 years agoBranch b1_5
yujian [Fri, 30 Jun 2006 13:42:17 +0000 (13:42 +0000)]
Branch b1_5
b=9853
Lustre cluster config scripts.
Move all the common variables and functions to lc_common.sh.
Modify Makefile.am to use *.sh.in to generate *.sh with @scriptlibdir@ replaced by the location of $(scriptlibdir).

18 years agob=4778
zhanghc [Fri, 30 Jun 2006 10:40:56 +0000 (10:40 +0000)]
b=4778
r=adilger
DESCRIPTION: fix a new bug caused by previous patch

18 years agoreply offset fix for EPROTO case in ptlrpc_connect_interpret().
lsy [Fri, 30 Jun 2006 02:13:05 +0000 (02:13 +0000)]
reply offset fix for EPROTO case in ptlrpc_connect_interpret().

18 years agoBranch b1_5
nathan [Thu, 29 Jun 2006 20:37:00 +0000 (20:37 +0000)]
Branch b1_5
b=8007
--writeconf only allowed a single target to regenerate it's log.

18 years agoDon't put "_lustre-b1_5.YYYYMMDDHHMM" in the extraversion as this kernel is
brian [Thu, 29 Jun 2006 16:25:24 +0000 (16:25 +0000)]
Don't put "_lustre-b1_5.YYYYMMDDHHMM" in the extraversion as this kernel is
not really at all lustre specific.

18 years agob=4778
zhanghc [Thu, 29 Jun 2006 10:45:45 +0000 (10:45 +0000)]
b=4778
r=adilger
DESCRIPTION: fix a bug

18 years agob=4778
zhanghc [Thu, 29 Jun 2006 04:28:30 +0000 (04:28 +0000)]
b=4778
r=dilger

DESCRIPTION:
    oscc->oscc_last_id and filter_last_id will mismatch
after using Lustre for a while, and will trigger LASSERTF.
this patch contains two modification:
    1, the whole operation of deleting orphans or precreating objects is
       protected by filter->fo_create_lock
    2, compare 'conn_cnt' before deleting or precreating, then avoid old
       request to be processed

18 years agoBranch b1_5
nathan [Wed, 28 Jun 2006 20:43:48 +0000 (20:43 +0000)]
Branch b1_5
b=8007
fix incorrect comment

18 years agoBranch b1_5
nathan [Wed, 28 Jun 2006 20:37:39 +0000 (20:37 +0000)]
Branch b1_5
b=8216
add help on --param
first whack at updating the lctl man page.

18 years agoBranch b1_5
nathan [Wed, 28 Jun 2006 17:29:24 +0000 (17:29 +0000)]
Branch b1_5
b=8007
fix tunefs.lustre -h

18 years agoBranch b1_5
nathan [Wed, 28 Jun 2006 17:16:18 +0000 (17:16 +0000)]
Branch b1_5
r=adilger
print lustre ver when obdclass module loads

18 years agofile 2.6-rhel4-titech.series was added on branch b1_5 on 2006-06-30 15:05:27 +0000
vitaly [Wed, 28 Jun 2006 10:59:45 +0000 (10:59 +0000)]
file 2.6-rhel4-titech.series was added on branch b1_5 on 2006-06-30 15:05:27 +0000

18 years agoDisable server building for patchless builds since patchless server is not
green [Wed, 28 Jun 2006 10:02:26 +0000 (10:02 +0000)]
Disable server building for patchless builds since patchless server is not
yet supported

18 years agoBranch b1_5
nathan [Wed, 28 Jun 2006 00:37:38 +0000 (00:37 +0000)]
Branch b1_5
b=9858
clean up loop dev if mount fails

18 years agob=8007
nathan [Wed, 28 Jun 2006 00:05:42 +0000 (00:05 +0000)]
b=8007
don't block other servers from starting when -o abort_recov is specified.

18 years agoBranch b1_5
nathan [Tue, 27 Jun 2006 22:23:23 +0000 (22:23 +0000)]
Branch b1_5
b=8007
-include luster vers in log markers
-fix missing index processing in tunefs.lustre

18 years agoBranch b1_5
scjody [Tue, 27 Jun 2006 18:12:56 +0000 (18:12 +0000)]
Branch b1_5
b=9387
r=adilger

Minor tweaks as part of the fix to 9387.

18 years agoAdd the right file for the patchless target.
brian [Tue, 27 Jun 2006 15:30:34 +0000 (15:30 +0000)]
Add the right file for the patchless target.

19 years agoremove redundant debug messages.
lsy [Tue, 27 Jun 2006 03:44:10 +0000 (03:44 +0000)]
remove redundant debug messages.

19 years agoLustre support needed to be able to build a 2.6 patchless kernel. This
brian [Tue, 27 Jun 2006 02:29:33 +0000 (02:29 +0000)]
Lustre support needed to be able to build a 2.6 patchless kernel.  This
simply builds a vanilla 2.6.16.22 kernel for the client nodes.

19 years agoBranch b1_5
nathan [Mon, 26 Jun 2006 23:03:43 +0000 (23:03 +0000)]
Branch b1_5
b=10670
if MODPROBE is set use kernel modules instead of local build modules (and
thus use modprobe.conf as well)

19 years agoBranch b1_5
nathan [Mon, 26 Jun 2006 20:47:34 +0000 (20:47 +0000)]
Branch b1_5
b=8007
change patch ver for beta 3

19 years agoBranch b1_5
nathan [Mon, 26 Jun 2006 19:24:16 +0000 (19:24 +0000)]
Branch b1_5
b=8007
get rid of trailing '/' at mount point, which confuses mtab

19 years agoBranch b1_5
nathan [Mon, 26 Jun 2006 18:08:16 +0000 (18:08 +0000)]
Branch b1_5
b=8007
enforce OST exclusivity with MDT,MGS

19 years agob=9863
yujian [Mon, 26 Jun 2006 10:07:42 +0000 (10:07 +0000)]
b=9863
Modify the condition to check whether a target was made to be high-available.

19 years agob=9853
yujian [Mon, 26 Jun 2006 10:04:42 +0000 (10:04 +0000)]
b=9853
1) Improve help infomation, add descriptions of each field in the csv format
2) Remove "heartbeat channels", "service address" and "heartbeat options" fields from the csv format. All the generated HA config files are suffixed with ".lustre".

19 years agob=9853
yujian [Mon, 26 Jun 2006 09:54:48 +0000 (09:54 +0000)]
b=9853
1) Add more checks and prompt messages before generating configuration files
2) Write Lustre HA information into *.lustre files instead of changing the existing HA configuration files

19 years agoBranch b1_5
nathan [Fri, 23 Jun 2006 19:41:48 +0000 (19:41 +0000)]
Branch b1_5
b=9858
more test fixes for b1_5

19 years agoBranch b1_5
nathan [Fri, 23 Jun 2006 19:02:32 +0000 (19:02 +0000)]
Branch b1_5
b=8007
ia64 doesn't have BLKGETSIZE64

19 years agoBranch b1_5
nathan [Fri, 23 Jun 2006 18:22:00 +0000 (18:22 +0000)]
Branch b1_5
b=8007
some better comments

19 years agoBranch b1_5
scjody [Fri, 23 Jun 2006 01:28:12 +0000 (01:28 +0000)]
Branch b1_5
b=9387
r=adilger

Select next connection according to last successful connection time.

19 years agoBranch b1_5
nathan [Fri, 23 Jun 2006 00:12:12 +0000 (00:12 +0000)]
Branch b1_5
b=9858
fix lots of cancel_lru_locks calls

19 years agoBranch b1_5
nathan [Thu, 22 Jun 2006 23:34:53 +0000 (23:34 +0000)]
Branch b1_5
b=9858
fix cancel_lru_locks()

19 years agoBranch b1_5
nathan [Thu, 22 Jun 2006 22:44:46 +0000 (22:44 +0000)]
Branch b1_5
b=6221
6221 didn't apply cleanly to b1_5

19 years agoBranch b1_5
nathan [Thu, 22 Jun 2006 22:44:09 +0000 (22:44 +0000)]
Branch b1_5
b=9858
remove all special mountconf test exceptions

19 years agoBranch b1_5
vitaly [Thu, 22 Jun 2006 22:09:31 +0000 (22:09 +0000)]
Branch b1_5

do not allow client_common_fill_super to LBUG with osfs.os_bsize == 0,
mfs_statfs returned 'cached' osfs value, when there was no a such.

19 years agoAC_CHECK_HEADERS([linux/xattr_acl.h]) check is incorrect and unused. Removing.
green [Thu, 22 Jun 2006 16:36:37 +0000 (16:36 +0000)]
AC_CHECK_HEADERS([linux/xattr_acl.h]) check is incorrect and unused. Removing.

19 years agominor fix of the offset for liblustre.
lsy [Thu, 22 Jun 2006 15:17:37 +0000 (15:17 +0000)]
minor fix of the offset for liblustre.

19 years agolinux/bit_spinlock.h is not present in some kernels. Introduce configure check
green [Thu, 22 Jun 2006 14:37:01 +0000 (14:37 +0000)]
linux/bit_spinlock.h is not present in some kernels. Introduce configure check

19 years agoBranch b1_5
vitaly [Thu, 22 Jun 2006 11:52:50 +0000 (11:52 +0000)]
Branch b1_5
b=10318

make new lfind behaviour the default one, only if some old (obsolete)
lfind option is specified, lfind works in the 'old' mode.

19 years agoMake patchless lustre client to compile and load on vanilla 2.6.17
green [Wed, 21 Jun 2006 21:48:09 +0000 (21:48 +0000)]
Make patchless lustre client to compile and load on vanilla 2.6.17

19 years agoMerge b1_5 from b1_4 (20060621_1641)
scjody [Wed, 21 Jun 2006 21:22:47 +0000 (21:22 +0000)]
Merge b1_5 from b1_4 (20060621_1641)

19 years agoMerge b1_5 from b1_4 (20060616_1515)
scjody [Wed, 21 Jun 2006 20:16:58 +0000 (20:16 +0000)]
Merge b1_5 from b1_4 (20060616_1515)

19 years agoBranch b1_5
nathan [Wed, 21 Jun 2006 16:59:51 +0000 (16:59 +0000)]
Branch b1_5
b=9860
a little safer and more verbose upgrade script

19 years agob=9863
yujian [Wed, 21 Jun 2006 10:48:16 +0000 (10:48 +0000)]
b=9863
1) Get "--stripe-count-hint" option
2) Calculate the inode size of mdt according to stripe count

19 years agoBranch b1_5
nathan [Tue, 20 Jun 2006 23:25:45 +0000 (23:25 +0000)]
Branch b1_5
b=8007
better leave this verbose until we get the kinks out

19 years agoBranch b1_5
nathan [Tue, 20 Jun 2006 21:42:51 +0000 (21:42 +0000)]
Branch b1_5
more info on err

19 years agoBranch b1_5
nathan [Tue, 20 Jun 2006 20:00:30 +0000 (20:00 +0000)]
Branch b1_5
b=8007
remove extra comma

19 years agoBranch b1_5
nathan [Tue, 20 Jun 2006 18:26:06 +0000 (18:26 +0000)]
Branch b1_5
b=8007
-tell users they may need our e2fsprogs
-tunefs.lustre must figure out if mgsnode is set

19 years ago- few more defines for liblustre to get updated ldlm buildable at least
alex [Tue, 20 Jun 2006 17:35:58 +0000 (17:35 +0000)]
- few more defines for liblustre to get updated ldlm buildable at least

19 years ago- missed to fix ns_lock in liblustre
alex [Tue, 20 Jun 2006 16:31:09 +0000 (16:31 +0000)]
- missed to fix ns_lock in liblustre

19 years agofile ext3-check-jbd-errors-2.6.9.patch was added on branch b1_5 on 2006-06-21 21...
scjody [Tue, 20 Jun 2006 15:53:46 +0000 (15:53 +0000)]
file ext3-check-jbd-errors-2.6.9.patch was added on branch b1_5 on 2006-06-21 21:22:45 +0000

19 years agofile ext3-check-jbd-errors-2.6.5.patch was added on branch b1_5 on 2006-06-21 21...
scjody [Tue, 20 Jun 2006 15:53:39 +0000 (15:53 +0000)]
file ext3-check-jbd-errors-2.6.5.patch was added on branch b1_5 on 2006-06-21 21:22:45 +0000

19 years agob=9332
alex [Tue, 20 Jun 2006 11:30:18 +0000 (11:30 +0000)]
b=9332

 - instead of taking a lock covering whole requested region ll_file_read()
   and ll_file_write() grab locks that cover a single stripe at most (but
   not in O_APPEND case). this way we improve overall stability because
   we don't hold locks awaiting some failed OST to recover.

19 years agob=9863
yujian [Tue, 20 Jun 2006 09:01:05 +0000 (09:01 +0000)]
b=9863
Change "failnode=" to "failover.node=" according to lustre_param.h

19 years agoNice catch Nathan. Forat difference between the three lines threw me for a
brian [Tue, 20 Jun 2006 00:53:46 +0000 (00:53 +0000)]
Nice catch Nathan.  Forat difference between the three lines threw me for a
loop, so fix that too.  :-)

19 years agoStay in parity with lustre_param.h, revision 1.1.4.5.
brian [Mon, 19 Jun 2006 23:59:11 +0000 (23:59 +0000)]
Stay in parity with lustre_param.h, revision 1.1.4.5.

19 years agob=9846
nathan [Mon, 19 Jun 2006 23:30:53 +0000 (23:30 +0000)]
b=9846
Change parameters to hierarchical format

19 years agoBranch b1_5
nathan [Mon, 19 Jun 2006 20:14:17 +0000 (20:14 +0000)]
Branch b1_5
b=8007
r=adilger
remove cruft in utils

19 years agoBranch b1_5
nathan [Mon, 19 Jun 2006 20:09:03 +0000 (20:09 +0000)]
Branch b1_5
b=8007
minor error message improvement

19 years agoBranch b1_5
nathan [Mon, 19 Jun 2006 19:20:51 +0000 (19:20 +0000)]
Branch b1_5
sector_t-overflow doesn't apply to my 2.6.10 fc3 uml

19 years agob=10088
alex [Mon, 19 Jun 2006 10:29:26 +0000 (10:29 +0000)]
b=10088

 - fine-grained locking for LDLM landed

19 years agoNeed to remember that these tests are not just done "in-tree", they are also
brian [Sat, 17 Jun 2006 22:05:32 +0000 (22:05 +0000)]
Need to remember that these tests are not just done "in-tree", they are also
done on properly installed (make install, rpm, etc.) lustre so...
When trying to install a modules, test if the module is in-tree and if not,
try to modprobe it.

19 years ago/mnt might not be on a writable filesystem, must $MOUNT should be, so just
brian [Sat, 17 Jun 2006 16:50:25 +0000 (16:50 +0000)]
/mnt might not be on a writable filesystem, must $MOUNT should be, so just
use its parent directory.

19 years agoCreate the mountpoints for the servers.
brian [Sat, 17 Jun 2006 12:20:07 +0000 (12:20 +0000)]
Create the mountpoints for the servers.

19 years agob=9853
yujian [Sat, 17 Jun 2006 07:14:08 +0000 (07:14 +0000)]
b=9853
Figure out the scripts path instead of hardcode it.

19 years agoAs per lustre_config.sh's new functionality, default the mount options to
brian [Fri, 16 Jun 2006 21:32:25 +0000 (21:32 +0000)]
As per lustre_config.sh's new functionality, default the mount options to
"noauto".

19 years agoThe mkfs_options are I guess more properly mkfs parameters and need to be
brian [Fri, 16 Jun 2006 20:27:51 +0000 (20:27 +0000)]
The mkfs_options are I guess more properly mkfs parameters and need to be
included, wrapped with --param="" in the format options.

19 years agomkfs options needs to be passed to mkfs.lustre inside of --param="". I
brian [Fri, 16 Jun 2006 19:31:34 +0000 (19:31 +0000)]
mkfs options needs to be passed to mkfs.lustre inside of --param="".  I
wonder if lustre_config.sh shouldn't do this wrapping for us before passing
to mkfs.lustre.

19 years agoVS: Branch b1_5
brian [Fri, 16 Jun 2006 18:57:41 +0000 (18:57 +0000)]
VS: Branch b1_5
Off-by-one in the OST entries.

19 years agoQuiet perl down a bit.
brian [Fri, 16 Jun 2006 17:29:16 +0000 (17:29 +0000)]
Quiet perl down a bit.

19 years agoAllow the caller to specify where the mounts on the nodes should be done.
brian [Fri, 16 Jun 2006 16:53:33 +0000 (16:53 +0000)]
Allow the caller to specify where the mounts on the nodes should be done.
Remove that silly "main()" function.

19 years agob=9853
yujian [Fri, 16 Jun 2006 16:06:27 +0000 (16:06 +0000)]
b=9853
1) Remove "-d opts" option, as mentioned in comment #96 of bug9853, using "mount options" field in the csv file to specify the options to be added into /etc/fstab.
2) Also, change the meaning of "-m" option to "don't modify /etc/fstab". If using this option, then the value of "mount options" field will be passed to mkfs.lustre instead of being added into /etc/fstab.

19 years agob=9863
yujian [Fri, 16 Jun 2006 13:53:40 +0000 (13:53 +0000)]
b=9863
"extents,mballoc" are default mount options of OST with ldiskfs.

19 years agoAfter adding the vim modeline into this file, the indentations of it became a bit...
yujian [Fri, 16 Jun 2006 10:38:56 +0000 (10:38 +0000)]
After adding the vim modeline into this file, the indentations of it became a bit mess due to my mix-using spaces and tabs originally. This version is neatly arranged.

19 years agoHave to include the contents of man_MANS in EXTRA_DIST, not just a reference
brian [Fri, 16 Jun 2006 00:29:07 +0000 (00:29 +0000)]
Have to include the contents of man_MANS in EXTRA_DIST, not just a reference
to it.  This is because man_MANS is conditional on building the utils but
even if the tree is configure without utils, it should not prevent the utils
files from being put in the "dist" tarball.

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 22:20:49 +0000 (22:20 +0000)]
Branch b1_5
b=9861
- turn on extents, mballoc on OSTs by default
- don't try to pre-check lustre/llite/ldiskfs options in mount.lustre.
Instead, just let the mount call fail.

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 20:55:21 +0000 (20:55 +0000)]
Branch b1_5
b=8216
put mount.lustre.8 back

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 20:51:37 +0000 (20:51 +0000)]
Branch b1_5
b=8216
add mount.lustre man page

19 years agoRemove mount.lustre.8 until the file shows up. Probably just a forgotten
brian [Thu, 15 Jun 2006 20:27:23 +0000 (20:27 +0000)]
Remove mount.lustre.8 until the file shows up.  Probably just a forgotten
cvs new on it, but this is breaking builds.

19 years agoQuick hack to be able to define the options in the /etc/fstab for the lustre
brian [Thu, 15 Jun 2006 19:40:43 +0000 (19:40 +0000)]
Quick hack to be able to define the options in the /etc/fstab for the lustre
filesystem added.
Probably some logic could be added in get_mntopts to detect the failover
option and only add noauto if it was not added by the command line argument.

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 18:52:20 +0000 (18:52 +0000)]
Branch b1_5
b=8216
man pages for mkfs.lustre, tunefs.lustre, mount.lustre, and Lustre

19 years agoDon't format devices. This hack will be replaced by a command line switch
brian [Thu, 15 Jun 2006 03:01:14 +0000 (03:01 +0000)]
Don't format devices.  This hack will be replaced by a command line switch
in lustre_config.sh to ask for a "no format, but configure" operation.

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 00:11:52 +0000 (00:11 +0000)]
Branch b1_5
b=8007
remove some non-applicable options from tunefs

19 years agoBranch b1_5
nathan [Thu, 15 Jun 2006 00:05:44 +0000 (00:05 +0000)]
Branch b1_5
b=8007
add a tunefs.lustre manpage

19 years agoBranch b1_5
nathan [Wed, 14 Jun 2006 23:28:53 +0000 (23:28 +0000)]
Branch b1_5
b=8007
oooh, a man page.

19 years agoThe start of a script to convert lmc "batch" specifications to the CSV
brian [Wed, 14 Jun 2006 22:03:42 +0000 (22:03 +0000)]
The start of a script to convert lmc "batch" specifications to the CSV
format used by the lustre_config.sh script to configure lustre.
The number of lmc options the script understands currently seems to cover
ltest's lmc batch generation.
This is the "cheap way" to get ltest to do mountconf and be able to do full
testing on b1_5 and has the added benefit of testing the lustre_config.sh
scripts while it's at it.
Note, that arguably we should be parsing the XML, not lmc batch commands.
Fortunately the way the script works, changing out parsing of lmc commands
for parsing XML is easy to do and will still work with the backend CSV
generation.

19 years agob=6062
green [Wed, 14 Jun 2006 20:57:09 +0000 (20:57 +0000)]
b=6062
r=adilger(original author)

SPEC SFS NFS v2 validation failure fix - make blocksize constant at 2*RPC size.

19 years agopatchless support.
green [Wed, 14 Jun 2006 20:26:10 +0000 (20:26 +0000)]
patchless support.

19 years agoBranch b1_5
nathan [Wed, 14 Jun 2006 19:55:38 +0000 (19:55 +0000)]
Branch b1_5
b=8007
mtab must be checked before writing anything.
we no longer need the loopback setup for reading.

19 years agoSome safety so that DEVICE_NAME[i] being "/dev/loop1" doesn't delete an
brian [Tue, 13 Jun 2006 19:05:14 +0000 (19:05 +0000)]
Some safety so that DEVICE_NAME[i] being "/dev/loop1" doesn't delete an
fstab entry with a device of "/dev/loop11".

19 years agoWhitespace consistency. I believe our standard is 4 spaces for indentation,
brian [Tue, 13 Jun 2006 15:49:29 +0000 (15:49 +0000)]
Whitespace consistency.  I believe our standard is 4 spaces for indentation,
no hard tabs.

19 years agob=9863
yujian [Tue, 13 Jun 2006 13:57:53 +0000 (13:57 +0000)]
b=9863
Use debugfs instead of mounting the target device to read the Lustre data files.

19 years agoMerge b1_5 from b1_4 (20060607_2142)
vitaly [Sun, 11 Jun 2006 16:54:36 +0000 (16:54 +0000)]
Merge b1_5 from b1_4 (20060607_2142)

19 years agob=9863
yujian [Sun, 11 Jun 2006 10:45:02 +0000 (10:45 +0000)]
b=9863
Add figure_journal_size() to get the default journal size of lustre target.

19 years agoBranch b1_5
nathan [Fri, 9 Jun 2006 23:11:05 +0000 (23:11 +0000)]
Branch b1_5
b=9860
better errors

19 years agoBranch b1_5
nathan [Fri, 9 Jun 2006 23:10:00 +0000 (23:10 +0000)]
Branch b1_5
b=9860
Add upgrade helper script per Andreas' suggestion

19 years agoBranch b1_5
nathan [Fri, 9 Jun 2006 19:29:43 +0000 (19:29 +0000)]
Branch b1_5
b=8007
Better error message for wrong fsname