View previous topic :: View next topic |
Author |
Message |
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Sun Aug 10, 2008 9:03 pm Post subject: |
|
|
dusanc wrote: |
I had a lot of hard reboots before I got UPS and stopped playing with kernel.
Reiser4 never let me down. By design it's atomic fs so eaither file is written or it's not.
|
Yeah, that's good in theory, but it's always nice to hear someone say it's actually working
dusanc wrote: |
I suggest r4+cryptocompress(LZO1 ofc) , and before backup do a fsck because it checksums your data (real checksumming, cc plugin does that). AFAIK r4 checksums while you use it.
|
Yeah, can't do a fsck on an online fs though, and I'd probably run daily backups overnight.
dusanc wrote: |
As for fsync heavy apps I'm not so sure. You may encounter slowdowns with that, but best would be to try R4.
|
Can't be worse than ext3 in that regard anyway.
dusanc wrote: | If you have a lot of big files (like movies and .iso) XFS is better AFAIK. |
Yeah I got a seperate drive for that, home $HOME and root is strictly small files.
dusanc wrote: |
I've just defragmented r4 after 1 year and can't notice any differences. Maybe because I'm using compression
Oh and loopback doesn't work for cryptocompress files, but works for ordinary unix plugin. There is a way to set your loopback file to unix plugin, search the reiserfs-devel m-l. |
Very nice, I didn't know that, then I wont worry about vmware/virtualbox failing on me.
Thanks |
|
Back to top |
|
|
dusanc Apprentice
Joined: 19 Sep 2005 Posts: 248 Location: Serbia
|
Posted: Mon Aug 11, 2008 11:10 am Post subject: |
|
|
Here's general algo how reiser4 plugins work wrt. checksumming:
Quote: | Kernel:
-----------
For unix-file plugin:
(1) set i_size and real size which coincide with each other;
(2) don't protect data by checksums (for performance reasons).
For cryptcompress plugin:
(1) set i_size which usually doesn't coincide with real size;
(2) don' t set real size for performance issues.
(3) set data checksums for every logical cluster;
|
http://marc.info/?l=reiserfs-devel&m=121796822223837&w=2 |
|
Back to top |
|
|
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Mon Aug 11, 2008 5:44 pm Post subject: |
|
|
Does anyone have the patch to silence reiser4.fsck's warnings on boot? It's a bit spammy, and I dont want to miss anything important. And running git fsck seems a bit scary. |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
|
Back to top |
|
|
Kluster n00b
Joined: 21 Jun 2005 Posts: 29
|
Posted: Mon Aug 11, 2008 9:40 pm Post subject: |
|
|
Wow that is a lot of votes for reiser4. I've been using reiser3 almost exclusively for years. I've had uncountable power outages while using reiserfs3 without ever a single incident of information loss. I always wonder if people are experiencing those types of issues due to usage or implementation.
Is reiser4's development still reasonable active? I really hope they follow Han's recommendations and that they let him work from prison.
Anyone try ext4 or zfs yet? When they solve their petty licensing issues and we are allow to implement ZFS in Linux without emulation, then I may give it a try. Usage in Linux software raid is becoming an issue for me.
For those of you interested in encryption Truecrypt is something to check out. All of the options so far seem to have default weaknesses, so it really asks a lot on the part of the user for a strong implementation and usage of the system. |
|
Back to top |
|
|
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Tue Aug 12, 2008 7:50 am Post subject: |
|
|
kernelOfTruth wrote: | neuron wrote: | Does anyone have the patch to silence reiser4.fsck's warnings on boot? It's a bit spammy, and I dont want to miss anything important. And running git fsck seems a bit scary. |
what warnings ? |
with cryptcompress fsck always shows errors, I've been told it'd be fixed in the new reiser4.fsck. |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Tue Aug 12, 2008 9:05 pm Post subject: |
|
|
I am using a bit of a bastard mix of filesystems at the moment. Ext2 for /boot reiser3 for / and /home, ext3 for a 1TB drive (i need files from it for windows) which actually a backup drive for a 1TB zfs-fuse pool made up of 2 500GB drives.
I've only had one bad experience with reiser3 which is when a partition, what i thought was a reiser partition, had bad superblocks. I used reiser tools on it and messed the whole thing up. Other than that it has been bomb proof. Don't like the FS check on mount though as it makes my boot sequence quite long.
zfs-fuse is pretty good. Only used it for a week now but i like the admin tools for it and how easy it is to use. It is not that fast to write too but read speeds are pretty good. Unfortunately the project looks orphaned and nobody in zfs knows how/wants to have the capability to expand raidz/2 pools, which is a pain. I also like how easy it is to enable compression and quotas[non-user] on any fs in a pool and where to set its mountpoint. |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
|
Back to top |
|
|
shaumux Veteran
Joined: 13 May 2005 Posts: 1009 Location: Hong Kong
|
Posted: Wed Aug 13, 2008 2:11 pm Post subject: |
|
|
Kluster wrote: | Wow that is a lot of votes for reiser4. I've been using reiser3 almost exclusively for years. I've had uncountable power outages while using reiserfs3 without ever a single incident of information loss. I always wonder if people are experiencing those types of issues due to usage or implementation. | I used reiser3 on my / partition uptill around 1.5 years back and then switched to reiser 4, never had any problems with it yet and with newer kernels it seems to be getting better to me.
As for ZFS i haven't used it but from the theoritical point it looks pretty good but btrfs seems even better and it seems to provide performance with features. |
|
Back to top |
|
|
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Wed Aug 13, 2008 2:33 pm Post subject: |
|
|
Put reiser4 on root, got patches from kernel.org, and latest in tree reiser4progs.
First sync'ed root over, that went fine (did it from a booted usb pen), then sync'ed my home directory, copied everything fine and umounted fine, but got errors on fsck requering --rebuild-fs, so I skipped using reiser4 on home for now.
Now while compiling stuff I noticed this in my dmesg:
Code: |
reiser4[rm(1789)]: reiser4_unlink_common (fs/reiser4/plugin/inode_ops.c:295)[nikita-3398]:
WARNING: Cannot unlink 802080 (-2)
|
A bit worried now I have to admit, computer survives overnight memtest, and overnight prime.
//edit *hardware error, bios upgrade made ram unstable, nothing to see here*
Last edited by neuron on Fri Aug 15, 2008 2:23 pm; edited 1 time in total |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
Posted: Wed Aug 13, 2008 3:10 pm Post subject: |
|
|
neuron wrote: | Put reiser4 on root, got patches from kernel.org, and latest in tree reiser4progs.
First sync'ed root over, that went fine (did it from a booted usb pen), then sync'ed my home directory, copied everything fine and umounted fine, but got errors on fsck requering --rebuild-fs, so I skipped using reiser4 on home for now.
Now while compiling stuff I noticed this in my dmesg:
Code: |
reiser4[rm(1789)]: reiser4_unlink_common (fs/reiser4/plugin/inode_ops.c:295)[nikita-3398]:
WARNING: Cannot unlink 802080 (-2)
|
A bit worried now I have to admit, computer survives overnight memtest, and overnight prime. |
what kernel & reiser4-patch are you using ?
you should at least use 2.6.26 + reiser4-patch _________________ https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa
Hardcore Gentoo Linux user since 2004 |
|
Back to top |
|
|
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Wed Aug 13, 2008 3:25 pm Post subject: |
|
|
kernelOfTruth wrote: | neuron wrote: | Put reiser4 on root, got patches from kernel.org, and latest in tree reiser4progs.
First sync'ed root over, that went fine (did it from a booted usb pen), then sync'ed my home directory, copied everything fine and umounted fine, but got errors on fsck requering --rebuild-fs, so I skipped using reiser4 on home for now.
Now while compiling stuff I noticed this in my dmesg:
Code: |
reiser4[rm(1789)]: reiser4_unlink_common (fs/reiser4/plugin/inode_ops.c:295)[nikita-3398]:
WARNING: Cannot unlink 802080 (-2)
|
A bit worried now I have to admit, computer survives overnight memtest, and overnight prime. |
what kernel & reiser4-patch are you using ?
you should at least use 2.6.26 + reiser4-patch |
I'm using 2.6.25-gentoo-r7 and http://www.kernel.org/pub/linux/kernel/people/edward/reiser4/reiser4-for-2.6/, which is marked for 2.6.25, so I figured that should be alright. |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Wed Aug 13, 2008 8:25 pm Post subject: |
|
|
Kluster wrote: |
As for ZFS i haven't used it but from the theoritical point it looks pretty good but btrfs seems even better and it seems to provide performance with features. |
I did use btrfs for a week but it is only for testing/review atm. Additionally i found it a little harder to work with (lack of documentation). |
|
Back to top |
|
|
shaumux Veteran
Joined: 13 May 2005 Posts: 1009 Location: Hong Kong
|
Posted: Wed Aug 13, 2008 9:46 pm Post subject: |
|
|
regomodo wrote: | I did use btrfs for a week but it is only for testing/review atm. Additionally i found it a little harder to work with (lack of documentation). | Well its not finished yet. Its not even at beta stage but it seems that it will be a rocking filesystem. |
|
Back to top |
|
|
dusanc Apprentice
Joined: 19 Sep 2005 Posts: 248 Location: Serbia
|
Posted: Wed Aug 13, 2008 9:53 pm Post subject: |
|
|
neuron wrote: | Put reiser4 on root, got patches from kernel.org, and latest in tree reiser4progs.
First sync'ed root over, that went fine (did it from a booted usb pen), then sync'ed my home directory, copied everything fine and umounted fine, but got errors on fsck requering --rebuild-fs, so I skipped using reiser4 on home for now.
Now while compiling stuff I noticed this in my dmesg:
Code: |
reiser4[rm(1789)]: reiser4_unlink_common (fs/reiser4/plugin/inode_ops.c:295)[nikita-3398]:
WARNING: Cannot unlink 802080 (-2)
|
A bit worried now I have to admit, computer survives overnight memtest, and overnight prime. |
That's a new one.
Could you please post a bugreport on reiserfs-devel@vger.kernel.org about your problem?
These things should be useful to know:
If you could remember what were those errors on fsck that required --rebuild-fs, post those too.
Also if you use some more layers like LVM, raid or so.
How much free space there is on that partition (in %)?
What command you used to create it and to mount it?
A way to reproduce it would be great.
Bad blocks?
I have the same setup (2.6.25-r7+reiser4-for-2.6.25.patch) so something is different.
That would be very helpful as this bug hadn't appeared earlier on m-l.
Thanks in advance
Dushan |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
Posted: Wed Aug 13, 2008 10:17 pm Post subject: |
|
|
dusanc wrote: | neuron wrote: | Put reiser4 on root, got patches from kernel.org, and latest in tree reiser4progs.
First sync'ed root over, that went fine (did it from a booted usb pen), then sync'ed my home directory, copied everything fine and umounted fine, but got errors on fsck requering --rebuild-fs, so I skipped using reiser4 on home for now.
Now while compiling stuff I noticed this in my dmesg:
Code: |
reiser4[rm(1789)]: reiser4_unlink_common (fs/reiser4/plugin/inode_ops.c:295)[nikita-3398]:
WARNING: Cannot unlink 802080 (-2)
|
A bit worried now I have to admit, computer survives overnight memtest, and overnight prime. |
That's a new one.
Could you please post a bugreport on reiserfs-devel@vger.kernel.org about your problem?
These things should be useful to know:
If you could remember what were those errors on fsck that required --rebuild-fs, post those too.
Also if you use some more layers like LVM, raid or so.
How much free space there is on that partition (in %)?
What command you used to create it and to mount it?
A way to reproduce it would be great.
Bad blocks?
I have the same setup (2.6.25-r7+reiser4-for-2.6.25.patch) so something is different.
That would be very helpful as this bug hadn't appeared earlier on m-l.
Thanks in advance
Dushan |
well, I apologize,
I was under the impression that this had already happened
@neuron:
what exactly were you trying to remove?
in the past there have already been cases where apps wouldn't want to end (kde4, bdc [bitdefender-console] to name a few)
no panic your data is safe ! (there has been no data corruption or loss so far)
that you encountered this minor "bug" (?) only means that reiser4 wasn't tested / used enough in this kind of usage-pattern & needs some optimization for this
if you provide enough information this is very likely to be fixed in no time
(<== this is just my view on the things, if I'm wrong or write BS, please correct me)
are you using default reiser4 or reiser4 with cryptcompress+lzo (recommended)
which i/o-scheduler are you using ? some VFS-settings to know would also be nice (specific settings / tweaks you made which might alter the behavior of the fs / VFS / data-handling)
sorry, if you already posted that ...
many thanks in advance
there the important code-bits (from latest reiser4, dunno if anything changed from .25)
Code: | /**
* reiser4_unlink_common - unlink of inode operations
* @parent: inode of directory to remove name from
* @victim: name to be removed
*
* This is common implementation of vfs's unlink method of struct
* inode_operations.
*/
int reiser4_unlink_common(struct inode *parent, struct dentry *victim)
{
reiser4_context *ctx;
int result;
struct inode *object;
file_plugin *fplug;
ctx = reiser4_init_context(parent->i_sb);
if (IS_ERR(ctx))
return PTR_ERR(ctx);
object = victim->d_inode;
fplug = inode_file_plugin(object);
assert("nikita-2882", fplug->detach != NULL);
result = unlink_check_areiser4_unlink_commonnd_grab(parent, victim);
if (result != 0) {
context_set_commit_async(ctx);
reiser4_exit_context(ctx);
return result;
}
result = fplug->detach(object, parent);
if (result == 0) {
dir_plugin *parent_dplug;
reiser4_dir_entry_desc entry;
parent_dplug = inode_dir_plugin(parent);
memset(&entry, 0, sizeof entry);
/* first, delete directory entry */
result = parent_dplug->rem_entry(parent, victim, &entry);
if (result == 0) {
/*
* if name was removed successfully, we _have_ to
* return 0 from this function, because upper level
* caller (vfs_{rmdir,unlink}) expect this.
*
* now that directory entry is removed, update
* stat-data
*/
reiser4_del_nlink(object, parent, 1);
/*
* Upon successful completion, unlink() shall mark for
* update the st_ctime and st_mtime fields of the
* parent directory. Also, if the file's link count is
* not 0, the st_ctime field of the file shall be
* marked for update. --SUS
*/
reiser4_update_dir(parent);
/* add safe-link for this file */
if (object->i_nlink == 0)
safe_link_add(object, SAFE_UNLINK);
}
}
if (unlikely(result != 0)) {
if (result != -ENOMEM)
warning("nikita-3398", "Cannot unlink %llu (%i)",
(unsigned long long)get_inode_oid(object),
result);
/* if operation failed commit pending inode modifications to
* the stat-data */
reiser4_update_sd(object);
reiser4_update_sd(parent);
}
reiser4_release_reserved(object->i_sb);
/* @object's i_ctime was updated by ->rem_link() method(). */
/* @victim can be already removed from the disk by this time. Inode is
then marked so that iput() wouldn't try to remove stat data. But
inode itself is still there.
*/
/*
* we cannot release directory semaphore here, because name has
* already been deleted, but dentry (@victim) still exists. Prevent
* balance_dirty_pages() from being called on exiting this context: we
* don't want to do this under directory i_mutex.
*/
context_set_commit_async(ctx);
reiser4_exit_context(ctx);
return result;
} |
_________________ https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa
Hardcore Gentoo Linux user since 2004 |
|
Back to top |
|
|
neuron Advocate
Joined: 28 May 2002 Posts: 2371
|
Posted: Thu Aug 14, 2008 8:05 am Post subject: |
|
|
I posted some details to the mailing list now, and I'm not overly worried about my data, I run consistent backups of everything worth keeping .
I haven't posted details about my rebuild-fs problem, and I wont until I can reproduce it. It was a problem I had when rsyncing data from one drive to another, while I was on a usb pen on console, so I got no logs/details. I will however see if I can reproduce it. |
|
Back to top |
|
|
jordanwb l33t
Joined: 10 Jul 2008 Posts: 642 Location: Ottawa, Canada
|
Posted: Thu Aug 21, 2008 7:03 pm Post subject: |
|
|
On my server I use XFS on top of a LVM2 array for easy enlargement of any filesystem without unmounting. I've tried ext3, Reiserfs and XFS; I found XFS to be the fastest as well.
On my laptop I used ext2 to reduce drive access, but today Kubuntu crashed and I couldn't fix it.
BTW for Xfs are there any mount flags that could improve performance? |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Fri Aug 22, 2008 7:31 pm Post subject: |
|
|
jordanwb wrote: | BTW for Xfs are there any mount flags that could improve performance? |
iirc setting "noatime" is quite critical for xfs. Or was reiserfs? |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
Posted: Fri Aug 22, 2008 7:36 pm Post subject: |
|
|
jordanwb wrote: | On my server I use XFS on top of a LVM2 array for easy enlargement of any filesystem without unmounting. I've tried ext3, Reiserfs and XFS; I found XFS to be the fastest as well.
On my laptop I used ext2 to reduce drive access, but today Kubuntu crashed and I couldn't fix it.
BTW for Xfs are there any mount flags that could improve performance? |
something along with the following kind
Code: | mount -o noatime,nodiratime,log bsize=256k,nobarrier,allocsize=512m -l internal,size=128m,logbufs=8 |
tuning-options during creation (an example):
Code: | mkfs.xfs -l size=128m,version=2 -d agcount=20 -n size=16k |
_________________ https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa
Hardcore Gentoo Linux user since 2004 |
|
Back to top |
|
|
jordanwb l33t
Joined: 10 Jul 2008 Posts: 642 Location: Ottawa, Canada
|
Posted: Fri Aug 22, 2008 8:38 pm Post subject: |
|
|
regomodo wrote: | jordanwb wrote: | BTW for Xfs are there any mount flags that could improve performance? |
iirc setting "noatime" is quite critical for xfs. Or was reiserfs? |
I already put noatime, but thanks anyways.
kernelOfTruth wrote: | Code: | mount -o noatime,nodiratime,log bsize=256k,nobarrier,allocsize=512m -l internal,size=128m,logbufs=8 |
|
What do the parameters after nodiratime do? How would I put that in the fstab file?
kernelOfTruth wrote: | Code: | mkfs.xfs -l size=128m,version=2 -d agcount=20 -n size=16k |
|
What do those do? |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Mon Sep 15, 2008 3:16 pm Post subject: |
|
|
I'm still using zfs-fuse and it is proving to be a reliable file-system so far. Also, it appears to not be an orphaned project. A new release show that people using 0.4.1 should really upgrade and that it has a few enhancements.
I'm tempted to put a new "bug" on bugzilla to get a version bump but i'm not too happy with that place. Get the feeling dev's dislike users. Until then i'm trying to make my own ebuild but simply changing version numbers and removing patch installs isn't enough.
Quote: | 12008-09-12 - Release 0.5.0
2--------------------------------------------------
3 * Updated ZFS code to Nevada build 98.
4Features:
5 * Implemented async I/O.
6 * gzip compression support (upstream).
7 * Filesystems can have multiple copies of data (zfs set fs copies=x) (upstream).
8 * zpool status -v now displays paths of files with permanent errors (upstream).
9 * Separate Intent Log (upstream, not tested!).
10 * Support for quotas and reservations that don't include snapshots and clones (upstream).
11 * Support for level-2 cache devices (upstream, not tested).
12 * Ported to the sparc64 architecture (arkadi).
13 * Implemented zfs send/recv.
14 * Turned zfs-fuse into a real daemon (Cameron Patrick, Bryan Donlan).
15Enhancements:
16 * Open block devices with O_EXCL.
17 * Log warning and error messages to syslog.
18Bug fixes:
19 * MAJOR: Made zfs-fuse flush the write cache of ATA/SATA/SCSI disks (Thanks Eric Anopolsky!).
20 * Fix possible corruption problem on 32-bit machines.
21 * Fix crash when using 'zpool status -v' to list errors.
22 * Allow suid files and device nodes to be stored on zfs (Cameron Patrick).
23 * Fix segfault when trying to run as non-root (Ming Zhang).
24 * Linux compatibility fix when trying to remove non-empty directory.
25 * Fix unexpected return codes when working with unlinked files/directories.
26 * Add support for gcc 4.2 (patch by Michal Kowalczuk, some fixes by Roland/devzero).
27 * Fixed bug where atime was not working.
28 * Fixed bug where "zpool create/add/attach" did not warn if device was part of existing pool. |
[edit] My ebuild almost works. It compiles fine (i had to add dev-libs/libaio to DEPEND). It fails at
Code: | >>> Install zfs-fuse-0.5.0 into /var/tmp/portage/sys-fs/zfs-fuse-0.5.0/image/ category sys-fs
cp: cannot stat `/usr/portage/local/myebuilds/sys-fs/zfs-fuse/files/zfs-fuse.rc-r1': No such file or directory | Not sure if i can figure this one out in the ebuild. |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Mon Sep 15, 2008 3:52 pm Post subject: |
|
|
Ok, it installs now. Made my backups, here comes the possibly painful reboot.
The build i used
Code: | # Copyright 1999-2007 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: /var/cvsroot/gentoo-x86/sys-fs/zfs-fuse/zfs-fuse-0.5.0.ebuild,v 1.1 2008/09/13 17:40:17 trapni Exp $
IUSE="doc debug"
inherit eutils
DESCRIPTION="An implementation of the ZFS filesystem for FUSE/Linux"
HOMEPAGE="http://www.wizy.org/wiki/ZFS_on_FUSE"
SRC_URI="http://download.berlios.de/zfs-fuse/zfs-fuse-${PV}.tar.bz2"
LICENSE="CDDL"
SLOT="0"
KEYWORDS="~x86 ~amd64"
DEPEND=">=sys-libs/glibc-2.3.3
>=dev-util/scons-0.96.1
>=sys-fs/fuse-2.6.1
dev-libs/libaio"
RDEPEND=">=sys-fs/fuse-2.6.1"
S=${WORKDIR}/${P}/src
src_unpack() {
unpack "${A}"
cd "${S}"
# epatch "${FILESDIR}/${P}-fix_zdb_path.patch"
# epatch "${FILESDIR}/${P}-fix_zfs-fuse_path.patch"
# epatch "${FILESDIR}/${P}-fix_ztest_path.patch"
}
src_compile() {
scons || die "Make failed"
}
src_install() {
if useq debug; then
mv cmd/ztest/ztest cmd/ztest/run-ztest || die
mv cmd/ztest/runtest.sh cmd/ztest/ztest || die
dosbin cmd/ztest/run-ztest || die
fi
dosbin cmd/ztest/ztest || die
if useq debug; then
mv zfs-fuse/zfs-fuse zfs-fuse/run-zfs-fuse || die
mv zfs-fuse/run.sh zfs-fuse/zfs-fuse || die
dobin zfs-fuse/run-zfs-fuse || die
fi
dosbin "zfs-fuse/zfs-fuse" || die
dosbin "cmd/zfs/zfs" || die
dosbin "cmd/zpool/zpool" || die
dosbin "cmd/zdb/zdb" || die
# newinitd "${FILESDIR}/zfs-fuse.rc-r1" "zfs-fuse" || die
cd "${WORKDIR}/${P}" || die
dodoc CHANGES || die
if use doc; then
dodoc {INSTALL,TODO,STATUS,TESTING,HACKING,BUGS} || die
fi
}
pkg_postinst() {
echo
einfo "To debug and play with ZFS-FUSE make sure you have a recent 2.6.xx"
einfo "series kernel with the FUSE module compiled in OR built as a"
einfo "kernel module."
einfo
einfo "You can start the ZFS-FUSE daemon by running"
einfo
einfo " /etc/init.d/zfs-fuse start"
einfo
einfo "as root from the command line. "
einfo
einfo "And don't forget to add it permanently, if you want to:"
einfo
einfo " rc-update add zfs-fuse boot"
einfo
einfo "For additional ZFS related commands I recommend the ZFS admin"
einfo "guide. http://opensolaris.org/os/community/zfs/docs/zfsadmin.pdf"
einfo
einfo "Don't forget this is an beta-quality release. Testing has been"
einfo "very limited so please make sure you backup any important data."
einfo
einfo "If you have any problems with zfs-fuse please visit the ZFS-FUSE."
einfo "website at http://developer.berlios.de/projects/zfs-fuse/"
echo
}
|
[edit] well, can still see my files and after a "zpool upgrade -v" i am told i'm using version 3. Umm? Is that right. I have no idea. |
|
Back to top |
|
|
regomodo Guru
Joined: 25 Mar 2008 Posts: 445
|
Posted: Wed Sep 17, 2008 11:50 am Post subject: |
|
|
Scratch that, it doesn't work. No idea. I don't know whether to give up or bite my lip and head on over to gentoo's bugzilla. |
|
Back to top |
|
|
Rouslan Nabioullin n00b
Joined: 22 Nov 2008 Posts: 36
|
Posted: Mon Dec 22, 2008 9:27 pm Post subject: Best Choice of Filesystems and Tune Options? |
|
|
I already installed Gentoo, but I was not satisfied with the filesystem performance (I only used ReiserFS). In addition, I installed an additional 250GB IDE HDD, so that provoked me to split my installation up into more partitions (for greater performance and other benefits). I have received very helpful suggestions from users on #gentoo, but I just want to see what you guys think and I'm sure there will be extra suggestions.
Proposed Config: (I adapted some parts)
Code: | HDD:
/boot ext2
swap (pri=2) sw
/tmp ext2
/usr/portage reiserfs
/home/admin/media xfs
SATA:
/ ext3
/var ext3
/usr ext3
swap (pri=1) sw
/home ext3
|
So far, I moved the installation to the IDE HDD and back to the SATA HDD, so I only have swap (pri=2), /tmp, /usr/portage, and /home/admin/media left to configure. In addition, I enabled full journaling (tune2fs -o journaldata) on the ext3 partitions, added the "relatime" mount option to all of the partitions (except swap, of course), and will probably increase the inode ratio later on.
The main thing I am concerned about is whether reiserfs would be appropriate for /usr/portage (since I was dissatisfied with its performance and am worried that perhaps system files became corrupted when I had used it on everything before), and if yes, whether I should add the relatime and/or notail mount option. I just want the best performance, a generous amount of inodes, and no corruption issues (I have a UPS connected). Thank you in advance for any suggestions (add/remove mount options, tune the filesystems, change the filesystems I haven't implemented yet, etc.).
Mod edit: Fixed the [code] tags for you. --timeBandit |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|