Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
>Many thanks for answering my question. Hopefully my noisy X4200 >will be installed in the data centre tomorrow (Thursday); I had >a set back today while fighting with the Remote Console feature >of ILOM 1.1.1 (i.e., it doesn't work). :-( Just ssh into it and use the serial console from within SSH. Casper ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: NFS share problem with mac os x client
The setup below works fine for me. macmini:~ jimb$ mount | grep jimb ride:/xraid2/home/jimb on /private/var/automount/home/jimb (nosuid, automounted) macmini:~ jimb$ nidump fstab / | grep jimb ride:/xraid2/home/jimb /home/jimb nfs rw,nosuid,tcp 0 0 NFS server: Solaris 10 11/06 x86_64 + patches, NFSv3. On the client there's latest release MacOSX version + patches. This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: zfs crashing
Same problem here after some patching :((( 42GB free in a 4.2TB zpool We can't upgrade to U3 without planning it. Is there any way to solve the problem?? remove latest patches? Our uptime with ZFS is going very low ... thanks Gino This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] NFS share problem with mac os x client
OSX *loves* NFS - it's a lot faster than Samba - but you need a bit of extra work. You need a user on the other end with the right uid and gid (assuming you're using NFSv3 - you probably are). Have a look at : http://number9.hellooperator.net/articles/2007/01/12/zfs-for-linux-and-osx-and-windows-and-bsd (especially the 'create a user' bit). On 07/02/07, Kevin Bortis <[EMAIL PROTECTED]> wrote: Hello, I test right now the beauty of zfs. I have installed opensolaris on a spare server to test nfs exports. After creating tank1 with zpool and a subfilesystem with zfs tank1/nfsshare, I have set the option sharenfs=on to tank1/nfsshare. With Mac OS X as client I can mount the filesystem in Finder.app with nfs://server/tank1/nfsshare, but if I copy a file an error ocours. Finder say "The operation cannot be completed because you do not have sufficient privileges for some of the items.". Until now I have shared the filesystems always with samba so I have almost no experience with nfs. Any ideas? Kevin This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss -- Rasputin :: Jack of All Trades - Master of Nuns http://number9.hellooperator.net/ ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Panic with "really out of space."
Hi All, I have a problem with Global Zone crashing along with the rest of the zones when a ZFS partition goes to 100% with the following funny messge (really out of space..) :-) It didn't leave the lying or joking out of space message though:-) Feb 8 10:39:35 ss44bsdvgza01 ^Mpanic[cpu1]/thread=2a100a43cc0: Feb 8 10:39:35 ss44bsdvgza01 unix: [ID 858701 kern.notice] really out of space Feb 8 10:41:03 ss44bsdvgza01 savecore: [ID 570001 auth.error] reboot after panic: really out of space Feb 8 10:41:03 ss44bsdvgza01 savecore: [ID 748169 auth.error] saving system crash dump in /var/crash/ss44bsdvzgza01/*.3 Checked on SunSolve and found that Update 3 might solve the problem. Is that the case ? Is there a patch coming out soon for this ? I don;'t think 124204-04 fix this problem. Correct me if I am wrong.. Thanks Roshan ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] ZFS multi-threading
With the CPU overhead imposed in checksum of blocks by ZFS, on a large sequential write test, the CPU was heavily loaded in a test that I ran. By turning off the checksum, the CPU load was greatly reduced. Obviously, this caused a tradeoff in reliability for CPU cycles. Would the logic behind ZFS take full advantage of a heavily multicored system, such as on the Sun Niagara platform? Would it utilize of the 32 concurrent threads for generating its checksums? Has anyone compared ZFS on a Sun Tx000, to that of a 2-4 thread x64 machine? This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] ZFS multi-threading
>With the CPU overhead imposed in checksum of blocks by ZFS, on a large >sequential write test, the CPU was heavily loaded in a test that I ran. >By turning off the checksum, the CPU load was greatly reduced. >Obviously, this caused a tradeoff in reliability for CPU cycles. What hardware platform and what was the I/O throughput at the peak and what was the difference in throughput and CPU utilization between both cases? Casper ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
Ivan Buetler wrote: Is this true for OpenSolaris? My experience: I was trying to upgrade from "SunOS 5.11 snv_28" to "SunOS 5.11 snv_54" where my NGZ zone roots were set to a zfs mount point like below: NAME USED AVAIL REFER MOUNTPOINT zpool 93.8G 40.1G26K /zpool zpool/zones 3.50G 40.1G 1.68G /zpool/zones Upgrading to SNV_54 did not work for me (CD|DVD|Live-Upgrade). The install procedure was cancelled after it came to the NGZ ZFS setup part. However - I was enforced to to a full re-install of the whole OS. By this time, I decided to have an OS independent application setup: I decided to leave all my Non-Solaris apps within the following structure: NAME USED AVAIL REFER MOUNTPOINT zpool 93.8G 40.1G26K /zpool zpool/applic2.40G 40.1G 2.40G /zpool/applic zpool/bin108M 40.1G 108M /zpool/bin zpool/data 644M 40.1G 644M /zpool/data zpool/logs 1.03G 40.1G 1.03G /zpool/logs This means, Apache, Tomcat, Bind DNS, Postfix, MySQL, Berkeley-DB, ... was installed using a prefix (e.g. ./configure --prefix=/zpool/applic/named) This gives me some independencies to the core OS located in /sbin; /usr/bin, ... After I moved all my apps into my own prefix path (ZFS mount poing), I did another full reinstall of the OS, where I found out that I should have backed up some files from the core OS before. Especially I should have backed up the following files from the GZ and all NGZ. a) /etc/hosts, /etc/passwd, /etc/shadow, /etc/nsswitch.conf, /etc/resolv.conf b) /etc/hostname.XX, c) /etc/init.d/startup-scripts (my own releases) After I did another full setup (not upgrading), I created the zones using the famous zonemgr script and brought back all applications by just mounting the /zpool/applic/path into the NGZ path. This way, I was pretty fast in upgrading the whole system to a new Nevada build, even upgrading would be the preferred solution to me. I do not know if I with SNV_54, another upgrade from SNV_54 to SNV_55 is supported by OpenSolaris. That is why this thread is of interest to me. Ivan, I am not sure if I completely understand your configuration, but you can upgrade a system with zones that have delegated zfs datasets or where you just used lofs mounts to mount the zfs filesystems into the zone. This would apply when all you have is data or non-Solaris pkgs installed in the zfs filesystems. Since the upgrade code does not have to discover and mount the zfs filesystems to perform the upgrade of the OS, this type of configuration works fine. We would have to see your actual zonecfg info to be sure that you haven't set things up in a way that would prevent the upgrade though. Jerry ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: NFS share problem with mac os x client
> In short - make sure your UID on Mac is enough to > access the files on > nfs (as it would be if you would try to access those > files locally). > Or perhaps you tried from user with uid=0 in which > case it's mapped to > nobody user by default. > > -- > Best regards, > Robert Exactly as Robert suggests. I've had the same "problem" but it turned out to simply be that I needed the OS X group and user ID's to match those setup on the ZFS OpenSolaris server. Once you correct this, the Finder will work great to a ZFS NFS share. This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
Ivan Buetler wrote: Jerry, Thank you for your response. See my zonecfg of the named NGZ here: [EMAIL PROTECTED] ~ # zonecfg -z named export create -b set zonepath=/zpool/zones/named set autoboot=true add inherit-pkg-dir set dir=/lib end add inherit-pkg-dir set dir=/platform end add inherit-pkg-dir set dir=/sbin end add inherit-pkg-dir set dir=/usr end add fs set dir=/zpool/applic/bind-9.3.2-P1 set special=/zpool/applic/bind-9.3.2-P1/ set type=lofs add options ro add options nodevices end add fs set dir=/zpool/data/named set special=/zpool/data/named set type=lofs add options rw add options nodevices end add net set address=1.2.3.4/27 set physical=qfe3 end add net set address=21.2.3.5/27 set physical=qfe3 end add net set address=10.10.10.10/24 set physical=qfe4 end add attr set name=comment set type=string set value="Zone named" end It looks like you won't be able to upgrade. Assuming /zpool is the mount of your zfs zpool, then your zonepath is on a zfs dataset so this is the exact issue that upgrade cannot handle yet. If you were to place your zones on a UFS filesystem then the other fs entries you have to mount zfs datasets within the zone would be fine. Jerry ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: ZFS on PC Based Hardware for NAS?
> I believe there is a write limit (commonly 10 > writes) on CF and > similar storage devices, but I don't know for sure. > Apart from that > I think it's a good idea. > > > James C. McPherson As a consequence, the /tmp, /var, and swap could eventually be moved to the ZFS hard drives to greatly reduce I/O to the CF card. This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: ZFS on PC Based Hardware for NAS?
[EMAIL PROTECTED] wrote on 02/08/2007 10:23:19 AM: > > I believe there is a write limit (commonly 10 > > writes) on CF and > > similar storage devices, but I don't know for sure. > > Apart from that > > I think it's a good idea. > > > > > > James C. McPherson > > As a consequence, the /tmp, /var, and swap could eventually be moved > to the ZFS hard drives to greatly reduce I/O to the CF card. > Or zfs's slab logic could have a randomized block selector where every write to the cf device gets written to a random free block instead of the disk based weighing that is done now. ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: ZFS on PC Based Hardware for NAS?
[EMAIL PROTECTED] wrote: I believe there is a write limit (commonly 10 writes) on CF and similar storage devices, but I don't know for sure. Apart from that I think it's a good idea. James C. McPherson As a consequence, the /tmp, /var, and swap could eventually be moved to the ZFS hard drives to greatly reduce I/O to the CF card. Or zfs's slab logic could have a randomized block selector where every write to the cf device gets written to a random free block instead of the disk based weighing that is done now. Most flash devices have wear leveling builtin, so I'm not sure that adding this feature to ZFS will accomplish much. IMHO, the more important ZFS feature, wrt flash, is COW. BTW, CFs can do a large number of random iops, much larger than any disk. -- richard ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: Advice on a cheap home NAS machine using ZFS
Dave Sneddon wrote: Can anyone shed any light on whether the actual software side of this can be achieved? Can I share my entire ZFS pool as a "folder" or "network drive" so WinXP can read it? Will this be fast enough to read/write to at DV speeds (25mbit/s)? Once the pool is set up and I have it shared within XP (assuming it can be done) can I then easily copy files to/from it? I've used Samba to share a Unix filesystem with Windows clients many times, and would recommend it for this project: http://www.samba.org It runs as a regular program on the Unix side, and does a very appropriate job of translating Unix filesystem semantics into Windows filesystem semantics. There are a few things that translate oddly -- Unix symlinks appear to be the real file on the Windows side, so one could really confuse a Windows user who isn't aware that they exist. Fortunately, most people who use symlinks know what they are, so I never had a problem with symlinks when I was running a 400-user NFS/Samba server (in my previous job). Also, traditional Unix permissions look funny if you try to adjust the permissions from a Windows workstation. And, lastly, usernames and passwords are hashed differently on Unix then they are on Windows, so you have to run smbpasswd on the Unix server before a particular user can access their files on the Unix server via Samba -- unless you configure Samba to play along with an existing Windows Domain or AD. All in all, Samba provides a nice bridge -- and I've found it to be worlds better than the Windows-based NFS clients, and more secure as well. I don't mind if I have to use something like FTP but ultimately it appearing as a drive in XP is my final goal. If this can't be done then I don't believe I will even attempt to install/create this server. Windows XP with the newer Office installations has a "web folders" facility, that kind-of-almost mounts an FTP server. It doesn't show up as a drive-letter, but it does appear in My Computer. It's not uncommon to run Samba, SSH/SFTP, and FTP servers on the same host -- though there is quite a lot to to be said for avoiding protocols such as plain old FTP where the login is sent across the network in plaintext. I hope this helps, -Luke smime.p7s Description: S/MIME Cryptographic Signature ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
On Thu, 8 Feb 2007, [EMAIL PROTECTED] wrote: > >Many thanks for answering my question. Hopefully my noisy X4200 > >will be installed in the data centre tomorrow (Thursday); I had > >a set back today while fighting with the Remote Console feature > >of ILOM 1.1.1 (i.e., it doesn't work). :-( > > Just ssh into it and use the serial console from within SSH. That's how I usually use the console on the X4200. However, that arrangement doesn't work when one wants to (re)install Solaris. Unless there's a way of telling the installer to use the serial console while booting from DVD, rather than using the GUI? -- Rich Teer, SCSA, SCNA, SCSECA, OpenSolaris CAB member President, Rite Online Inc. Voice: +1 (250) 979-1638 URL: http://www.rite-group.com/rich ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Peculiar behaviour of snapshot after zfs receive
I am seeing what I think is very peculiar behaviour of ZFS after sending a full stream to a remote host - the upshot being that I can't send an incremental stream afterwards. What I did was this: host1 is Solaris 10 Update 2 SPARC host2 is Solaris 10 Update 2 x86 host1 # zfs snapshot work/[EMAIL PROTECTED] host1 # zfs send work/[EMAIL PROTECTED] | ssh host2 zfs recv export/home host1 # ssh host2 host2 # zfs list export/home 1.02G 47.8G 1.02G /export/home export/[EMAIL PROTECTED] 70.5K - 1.02G - host2 # Note that the snapshot on the remote system is showing changes to the underlying filesystem, even though it is not accessed by any application on host2. Now, I try to send an incremental stream: host1 # zfs snapshot work/[EMAIL PROTECTED] host1 # zfs send -i work/[EMAIL PROTECTED] work/[EMAIL PROTECTED] | ssh host2 zfs recv export/home cannot receive: destination has been modified since most recent snapshot -- use 'zfs rollback' to discard changes Am I using send/recv incorrectly or is there something else going on here that I am missing? Thanks, Trev smime.p7s Description: S/MIME Cryptographic Signature ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] update of zfs boot support
We've gotten a lot of questions lately about when we'll have an updated version of support for booting from zfs. We are aiming at a new version of this going in to build 60. New instructions for setting up this configuration will be made available at the same time. If build 60 turns out not to be possible, I'll notify this mailing list. But for now, build 60 is our target. This update will still be for x86 platforms only. Lori ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Peculiar behaviour of snapshot after zfs receive
Hello Trevor, Thursday, February 8, 2007, 6:23:21 PM, you wrote: TW> I am seeing what I think is very peculiar behaviour of ZFS after sending a TW> full stream to a remote host - the upshot being that I can't send an TW> incremental stream afterwards. TW> What I did was this: TW> host1 is Solaris 10 Update 2 SPARC TW> host2 is Solaris 10 Update 2 x86 TW> host1 # zfs snapshot work/[EMAIL PROTECTED] TW> host1 # zfs send work/[EMAIL PROTECTED] | ssh host2 zfs recv export/home TW> host1 # ssh host2 TW> host2 # zfs list TW> export/home 1.02G 47.8G 1.02G /export/home TW> export/[EMAIL PROTECTED] 70.5K - 1.02G - TW> host2 # TW> Note that the snapshot on the remote system is showing changes to the TW> underlying filesystem, even though it is not accessed by any application on host2. TW> Now, I try to send an incremental stream: TW> host1 # zfs snapshot work/[EMAIL PROTECTED] TW> host1 # zfs send -i work/[EMAIL PROTECTED] work/[EMAIL PROTECTED] | ssh host2 zfs recv TW> export/home TW> cannot receive: destination has been modified since most recent snapshot -- TW> use 'zfs rollback' to discard changes TW> Am I using send/recv incorrectly or is there something else going on here that TW> I am missing? It's a known bug. umount and rollback file system on host 2. You should see 0 used space on a snapshot and then it should work. -- Best regards, Robertmailto:[EMAIL PROTECTED] http://milek.blogspot.com ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Peculiar behavior of snapshot after zfs receive
> > TW> Am I using send/recv incorrectly or is there something else > going on here that > TW> I am missing? > > > It's a known bug. > > umount and rollback file system on host 2. You should see 0 used space > on a snapshot and then it should work. Bug ID? Is it related to atime changes? > > -- > Best regards, > Robertmailto:[EMAIL PROTECTED] >http://milek.blogspot.com ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re[2]: [zfs-discuss] ZFS Degraded Disks
Hello Kory, Thursday, February 8, 2007, 12:33:13 AM, you wrote: KW> I run the ZFS command and get this below. How do you fix a degraded KW> disk? KW> zpool replace moodle c1t3d0 KW> invalid vdev specification KW> use '-f' to override the following errors: KW> /dev/dsk/c1t3d0s0 is part of active ZFS pool moodle. Please see zpool(1M). KW> /dev/dsk/c1t3d0s2 is part of active ZFS pool moodle. Please see zpool(1M). zpool status output please first -- Best regards, Robertmailto:[EMAIL PROTECTED] http://milek.blogspot.com ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
>That's how I usually use the console on the X4200. However, that >arrangement doesn't work when one wants to (re)install Solaris. >Unless there's a way of telling the installer to use the serial >console while booting from DVD, rather than using the GUI? I thought there were a grub "use ttya" and "use ttyb" line on the DVD? Casper ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
On Thu, 8 Feb 2007, [EMAIL PROTECTED] wrote: > I thought there were a grub "use ttya" and "use ttyb" line on the DVD? Yes but one needs to be able to see that menu in order to select the correct item first. A chicken-and-egg situation! Not that it matters so much for this case now, as I've hooked up a spare monitor and keyboard to it. But connecting a monitor and keyboard directly to a server just feels ... wrong. But then I'm an old-school SPARC guy, so I guess that's not too surprising! :-) -- Rich Teer, SCSA, SCNA, SCSECA, OpenSolaris CAB member President, Rite Online Inc. Voice: +1 (250) 979-1638 URL: http://www.rite-group.com/rich ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zones-discuss] Re: [zfs-discuss] Downsides to zone roots on ZFS?
>Yes but one needs to be able to see that menu in order to select >the correct item first. A chicken-and-egg situation! But the console redirection setting in the BIOS should address that, right? Casper ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re[2]: [zfs-discuss] Peculiar behavior of snapshot after zfs receive
Hello Wade, Thursday, February 8, 2007, 8:00:40 PM, you wrote: >> >> TW> Am I using send/recv incorrectly or is there something else >> going on here that >> TW> I am missing? >> >> >> It's a known bug. >> >> umount and rollback file system on host 2. You should see 0 used space >> on a snapshot and then it should work. WSfc> Bug ID? Is it related to atime changes? It has to do with delete queue being processed when fs is mounted. The bug id is: 6343779 http://bugs.opensolaris.org/view_bug.do?bug_id=6343779 -- Best regards, Robertmailto:[EMAIL PROTECTED] http://milek.blogspot.com ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Peculiar behaviour of snapshot after zfs receive
On Feb 8, 2007, at 10:53 AM, Robert Milkowski wrote: Hello Trevor, Thursday, February 8, 2007, 6:23:21 PM, you wrote: TW> I am seeing what I think is very peculiar behaviour of ZFS after sending a TW> full stream to a remote host - the upshot being that I can't send an TW> incremental stream afterwards. TW> What I did was this: TW> host1 is Solaris 10 Update 2 SPARC TW> host2 is Solaris 10 Update 2 x86 TW> host1 # zfs snapshot work/[EMAIL PROTECTED] TW> host1 # zfs send work/[EMAIL PROTECTED] | ssh host2 zfs recv export/home TW> host1 # ssh host2 TW> host2 # zfs list TW> export/home 1.02G 47.8G 1.02G /export/home TW> export/[EMAIL PROTECTED] 70.5K - 1.02G - TW> host2 # TW> Note that the snapshot on the remote system is showing changes to the TW> underlying filesystem, even though it is not accessed by any application on host2. TW> Now, I try to send an incremental stream: TW> host1 # zfs snapshot work/[EMAIL PROTECTED] TW> host1 # zfs send -i work/[EMAIL PROTECTED] work/[EMAIL PROTECTED] | ssh host2 zfs recv TW> export/home TW> cannot receive: destination has been modified since most recent snapshot -- TW> use 'zfs rollback' to discard changes TW> Am I using send/recv incorrectly or is there something else going on here that TW> I am missing? It's a known bug. umount and rollback file system on host 2. You should see 0 used space on a snapshot and then it should work. And with snv_48 (s10u4 when it becomes available), you can use 'zfs recv -F' to force the rollback. eric ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: Re[2]: [zfs-discuss] Peculiar behavior of snapshot after zfs receive
> Hello Wade, > > Thursday, February 8, 2007, 8:00:40 PM, you wrote: > > > > > >> > >> TW> Am I using send/recv incorrectly or is there something else > >> going on here that > >> TW> I am missing? > >> > >> > >> It's a known bug. > >> > >> umount and rollback file system on host 2. You should see 0 used space > >> on a snapshot and then it should work. > > WSfc> Bug ID? Is it related to atime changes? > > It has to do with delete queue being processed when fs is mounted. > > > The bug id is: 6343779 > http://bugs.opensolaris.org/view_bug.do?bug_id=6343779 > Robert, Thanks! This is good to know, I was having issues with one of my boxes and zfs send/recive that may very well have been this bug. -Wade ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: The ZFS MOS and how DNODES are stored
Bill Moloney wrote: Thanks for the input Darren, but I'm still confused about DNODE atomicity ... it's difficult to imagine that a change that is made anyplace in the zpool would require copy operations all the way back up to the uberblock This is in fact what happens. However, these changes are all batched up (into a transaction group, or "txg"), so the overhead is minimal. > the DNODE implementation appears to include its own checksum field (self-checksumming), That is not the case. Only the uberblock and intent log blocks are self-checksumming. > if this is not the case, than 'any' modification in the zpool would require copying up to the uberblock That's correct, any modifications require modifying the uberblock (with the exception of intent log writes). FYI, dnodes are not involved with the snapshot mechanism. Snapshotting happens at the dsl dataset layer, while dnodes are implemented above that in the dmu layer. Check out dsl_dataset.[ch]. --matt ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] ZFS multi-threading
> Would the logic behind ZFS take full advantage of a heavily multicored > system, such as on the Sun Niagara platform? Would it utilize of the > 32 concurrent threads for generating its checksums? Has anyone > compared ZFS on a Sun Tx000, to that of a 2-4 thread x64 machine? Pete and I are working on resolving ZFS scalability issues with Niagara and StarCat right now. I'm not sure if any official numbers about ZFS performance on Niagara have been published. As far as concurrent threads generating checksums goes, the system doesn't work quite the way you have postulated. The checksum is generated in the ZIO_STAGE_CHECKSUM_GENERATE pipeline state for writes, and verified in the ZIO_STAGE_CHECKSUM_VERIFY pipeline stage for reads. Whichever thread happens to advance the pipline to the checksum generate stage is the thread that will actually perform the work. ZFS does not break the work of the checksum into chunks and have multiple CPUs perform the computation. However, it is possible to have concurrent writes simultaneously in the checksum_generate stage. More details about this can be found in zfs/zio.c and zfs/sys/zio_impl.h -j ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Overview (rollup) of recent activity on zfs-discuss
For background on what this is, see: http://www.opensolaris.org/jive/message.jspa?messageID=24416#24416 http://www.opensolaris.org/jive/message.jspa?messageID=25200#25200 = zfs-discuss 01/16 - 01/31 = Size of all threads during period: Thread size Topic --- - 76 Thumper Origins Q 68 ZFS or UFS - what to do? 52 How much do we really want zpool remove? 42 Heavy writes freezing system 28 can I use zfs on just a partition? 56 External drive enclosures + Sun Server for massstorage 16 hot spares - in standby? 16 Cheap ZFS homeserver. 15 panic with zfs 12 multihosted ZFS 11 zpool split 10 zfs rewrite? 10 What SATA controllers are people using for ZFS? 10 UFS on zvol: volblocksize and maxcontig 10 Adding my own compression to zfs 9 Synchronous Mount? 8 External drive enclosures + Sun 7 zpool dumps core with did device 7 ZFS inode equivalent 7 ZFS direct IO 7 External drive enclosures + Sun Server for mass 6 need advice: ZFS config ideas for X4500 Thumper? 6 file not persistent after node bounce when there is a bad disk? 6 Project Proposal: Availability Suite 6 External drive enclosures + Sun Server for 6 Backup/Restore idea? 5 zfs / nfs issue (not performance :-) with courier-imap 5 restore pool from detached disk from mirror 5 high density SAS 5 ditto==RAID1, parity==RAID5? 5 ZFS and HDLM 5.8 ... does that coexist well ? 5 Solaris-Supported cards with battery backup 5 I only see 5.33TB of my 7.25TB zfs-pool. Why? 5 Can you turn on zfs compression when the fs is already populated? 4 dumpadm and using dumpfile on zfs? 4 ZFS patches for Solaris 10U2 ? 4 ZFS panics system during boot, after 11/06 upgrade 4 ZFS on PC Based Hardware for NAS? 4 ZFS brings System to panic/freeze 4 Some questions I had while testing ZFS. 4 On-failure policies for pools 4 Mounting a ZFS clone 4 Implementation Question 4 Folders vs. ZFS 4 Export ZFS over NFS ? 3 zfs crashing 3 periodic disk i/o upon pool upgrade 3 patch making tar multi-thread 3 iSCSI on a single interface? 3 data wanted: disk kstats 3 bug id 6381203 3 Need Help on device structure 3 How to reconfigure ZFS? 3 Actual (cache) memory use of ZFS? 3 A little different look at filesystems ... Just looking for ideas 2 zpool overlay 2 unable to boot zone 2 question: zfs code size statistics 2 bug id 6343667 2 ZFS volume is hosing BIOS POST on Ultra20 (BIOS 2.1.7) 2 ZFS block squashing (CAS) 2 ZFS and HDLM 5.8 ... does that coexist well ? [MD21] 2 X2100 not hotswap 2 Why replacing a drive generates writes to other disks? 2 SAS support on Solaris 2 Problems adding drive 2 Extremely poor ZFS perf and other observations 2 Enhance 1U eSATA storage device and Solaris 10? 1 yet another blog: ZFS space, performance, MTTDL 1 unsubscribe 1 question about self healing 1 ftruncate is failing on ZFS 1 Zpooling problems 1 ZFSroot hanging at boot time with os nv54 1 ZFS ARC blog 1 VxVM volumes in a zpool. 1 Understanding ::memstat in terms of the ARC 1 Remote Replication 1 Raid Edition drive with RAIDZ 1 Possibility to change GUID zfs pool at import 1 On the SATA framework 1 Multiple Read one Writer Filesystem 1 MTTDL blogfest continues 1 HELP please zfs can't open drives!!! 1 Eliminating double path with ZFS's volume manager 1 Distributed FS 1 Converting home directory from ufs to zfs 1 Bathing ape hoody Bathing ape bape hoodie lil wayne BBC 1 Almost lost my data 1 A little different look at filesystems ... Justlooking for ideas Posting activity by person for period: # of posts By -- -- 38 rmilkowski at task.gda.pl (robert milkowski) 38 fcusack at fcusack.com (frank cusack) 33 jasonjwwilliams at gmail.com (jason j. w. williams) 26 rheilke at dragonhearth.com (rainer heilke) 25 r
[zfs-discuss] FROSUG February Meeting Announcement (2/22/2007)
This month's FROSUG (Front Range OpenSolaris User Group) meeting is on Thursday, February 22, 2007. Our presentation is "ZFS as a Root File System" by Lori Alt. In addition, Jon Bowman will be giving an OpenSolaris Update, and we will also be doing an InstallFest. So, if you want help installing an OpenSolaris distribution, backup your laptop and bring it to the meeting! About the presentation(s): One of the next steps in the evolution of ZFS is to enable its use as a root file system. This presentation will focus on how booting from ZFS will work, how installation will be affected by ZFS's feature set, and the many advantages that will result from being able to use ZFS as a root file system. The presentation(s)s will be posted here prior to the meeting: http://www.opensolaris.org/os/community/os_user_groups/frosug/ About our presenter(s): Lori Alt is a Staff Engineer at Sun Microsystems, where she has worked since 1991. Lori worked on Solaris install and upgrade and then on UFS, where she led the multi-terabyte UFS project. She has Bachelor's and Master's degrees in computer science from Washington University in St. Louis, MO. - Meeting Details: When: Thursday, February 22, 2007 Times: 6:00pm - 6:30pm Doors open and Pizza 6:30pm - 6:45pm OpenSolaris Update (Jon Bowman) 6:45pm - 8:30pm ZFS as a Root File System (Lori Alt) Where: Sun Broomfield Campus Building 1 - Conference Center 500 Eldorado Blvd. Broomfield, CO 80021 Note: The location of this meeting may change. We will send out an additional email prior to the meeting if this happens. Pizza and soft drinks will be served at the beginning of the meeting. Please RSVP to frosug-rsvp(AT)opensolaris(DOT)org in order to help us plan for food and setup access to the Sun campus. We hope to see you there! Thanks, FROSUG +++ Future Meeting Plans: March 29, 2007: Doug McCallum presents "sharemgr" This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
