Re: [zfs-discuss] j4200 drive carriers
nevermind, i will just get a Promise array. Don't. I don't normally like to badmouth vendors, but my experience with Promise was one of the worst in my career, for reasons that should be relevant other ZFS-oriented customers. recommendations for an alternative? ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Oracle raw volumes
one question related with this: would KAIO be supported on such configuration ? At least trying to open /dev/zvol/rdsk/datapool/master where master is defined as: # zpool create -f datapool mirror c1t1d0 c2t0d0 # zfs create -V 1gb datapool/master # ls -lrt /dev/zvol/rdsk/datapool/master lrwxrwxrwx 1 root root 39 Feb 1 18:29 /dev/zvol/rdsk/datapool/master - ../../../../devices/pseudo/z...@0:3c,raw resulted with: # truss -t kaio,lwp_create ./test-kaio /dev/zvol/rdsk/datapool/master kaio(AIOINIT) = 0 lwp_create(0x08046860, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804685C) = 2 /2: lwp_create() (returning as new lwp ...) = 0 /1: kaio(AIOREAD, 5, 0x08046C2C, 1024, 0, 0x0804702C) Err#81 EBADFD /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 3 /3: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 4 /4: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 5 /5: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 6 /6: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 7 /7: lwp_create() (returning as new lwp ...) = 0 /4: kaio(AIONOTIFY) = 0 /1: kaio(AIOWAIT, 0x, 0) = 1 aio succeeded Comments ? The original question came from a real setup using Sybase which spits a lot of: 733/1: 4137604 6 1 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137611 7 3 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 1 0 733/1: 4137614 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137622 5 2 recv(0x17, 0x100044F20F0, 0x800) = 109 0 733/1: 4137626 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137631 6 2 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137633 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137636 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137637 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137641 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137642 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137646 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137647 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137650 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137652 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137655 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137657 4 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137660 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 Thanks, Stefan test-kaio.c http://sunsite.uakom.sk/sunworldonline/swol-07-1998/swol-07-insidesolaris.html -- 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] j4200 drive carriers
On Sun, Feb 1, 2009 at 10:30 AM, Frank Cusack fcus...@fcusack.com wrote: nevermind, i will just get a Promise array. Don't. I don't normally like to badmouth vendors, but my experience with Promise was one of the worst in my career, for reasons that should be relevant other ZFS-oriented customers. recommendations for an alternative? I would imagine something like this: http://www.pc-pitstop.com/sas_cables_enclosures/scsase16.asp or this: http://www.pc-pitstop.com/sas_cables_enclosures/scsas16rm.asp would get the job don. ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Zpool corrupt, crashes system when imported -- one last try
Sorry for following up to myself. It was suggested off-list that I try to see if the slightly older code in the Belenix liveCD would maybe work with my pool. Nope -- because I upgraded my pool back when I thought OpenSolaris 2008.11 was going to work, so it's too new a version for the code on that liveCD. I *did* try the 2008.11 liveCD, in addition to my installation, and they both crash when I import the pool. And I'm not going to do anything drastic and permanent today after all. Hoping for insight from somebody! On Sat, January 31, 2009 22:51, David Dyer-Bennet wrote: As I've said, I've got this ZFS pool, with 450GB of my data in it, that crashes the system when I import it. I've now got a dozen or so log entries on this, having finally gotten it to happen in a controlled enough environment that I can get at the log files and transfer them somewhere I can access them. They all appear to be exactly the same thing. This crash happens nearly instantly after I do a zpool import -f zp1. Since the stack traceback suggests it's involved in a scrub, and indeed I do think it was scrubbing when this first started happening, I've tried doing zpool import -f zp1; zpool scrub -s zp1 in hopes that it will get in in time to stop the hypothetical scrub. I'm running OpenSolaris 2008.11. I'd like to rescue this pool. If I can't, though, I've got to destroy it and restore from backup (which shouldn't make things much worse than they already are; my backups are in decent shape so far as I know). And I've spent far far too much time on this (busy at work, other things at home, so this has dragged on interminably). And I'd love to provide useful information to anybody interested in finding and fixing whatever is wrong in the code that left me in this position, of course. Nobody has responded to my image of the stack traceback from yesterday, but I'm hoping that now that I've managed to get more information (including the stuff before the stack traceback), somebody may be able to do something with it. I do also seem to have a dump, if that's any use to anybody: Jan 31 22:34:11 fsfs genunix: [ID 111219 kern.notice] dumping to /dev/zvol/dsk/rpool/dump, offset 65536, content: kernel Jan 31 22:34:21 fsfs genunix: [ID 409368 kern.notice] ^M100% done: 197356 pages dumped, compression ratio 3.20, Jan 31 22:34:21 fsfs genunix: [ID 851671 kern.notice] dump succeeded Here's how the pool shows before I try to import it: local...@fsfs:~$ pfexec zpool import pool: zp1 id: 4278074723887284013 state: ONLINE action: The pool can be imported using its name or numeric identifier. config: zp1 ONLINE mirrorONLINE c5t0d0 ONLINE c5t1d0 ONLINE mirrorONLINE c6t0d0 ONLINE c6t1d0 ONLINE local...@fsfs:~$ Should that report say if a scrub was in progress? And here's the crash that happens immediately when I do import it: Jan 31 22:34:10 fsfs unix: [ID 836849 kern.notice] Jan 31 22:34:10 fsfs ^Mpanic[cpu1]/thread=ff00045fdc80: Jan 31 22:34:10 fsfs genunix: [ID 335743 kern.notice] BAD TRAP: type=e (#pf Page fault) rp=ff00045fcd60 addr=4e8 occurred in module unix due to a NULL pointer dereference Jan 31 22:34:10 fsfs unix: [ID 10 kern.notice] Jan 31 22:34:10 fsfs unix: [ID 839527 kern.notice] sched: Jan 31 22:34:10 fsfs unix: [ID 753105 kern.notice] #pf Page fault Jan 31 22:34:10 fsfs unix: [ID 532287 kern.notice] Bad kernel fault at addr=0x4e8 Jan 31 22:34:10 fsfs unix: [ID 243837 kern.notice] pid=0, pc=0xfb84e84b, sp=0xff00045fce58, eflags=0x10246 Jan 31 22:34:10 fsfs unix: [ID 211416 kern.notice] cr0: 8005003bpg,wp,ne,et,ts,mp,pe cr4: 6f8xmme,fxsr,pge,mce,pae,pse,de Jan 31 22:34:10 fsfs unix: [ID 624947 kern.notice] cr2: 4e8 Jan 31 22:34:10 fsfs unix: [ID 625075 kern.notice] cr3: 340 Jan 31 22:34:10 fsfs unix: [ID 625715 kern.notice] cr8: c Jan 31 22:34:10 fsfs unix: [ID 10 kern.notice] Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]rdi: 4e8 rsi: a200 rdx: ff00045fdc80 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]rcx:1 r8: ff01599c8540 r9:0 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]rax:0 rbx:0 rbp: ff00045fced0 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]r10: 5b40 r11:0 r12: ff0161e92040 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]r13: ff0176684800 r14: ff0161e92040 r15:0 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]fsb:0 gsb: ff0149f68500 ds: 4b Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice] es: 4b fs:0 gs: 1c3 Jan 31 22:34:10 fsfs unix: [ID 592667 kern.notice]trp:e err:2 rip:
Re: [zfs-discuss] j4200 drive carriers
On Sun, Feb 1, 2009 at 11:30, Frank Cusack fcus...@fcusack.com wrote: recommendations for an alternative? At work we just ordered a Supermicro CSE-846E1-R900B [1]: 24 hot-swap bays, redundant 900W power supplies, LSI SAS expander. This doesn't quite make a JBOD by itself (it's designed to be used as a case for a whole system); you need to add a PCI-slot-based external-to-internal SAS cable [2] and a so-called JBOD control board which makes the case power up without a motherboard in it. This is the cheapest SAS expander-based chassis I've found; even after adding the two additional parts mentioned above, cost from provantage.com shipped is around $1250 (depending on your location, of course). You'll need to add a SAS controller of your choice and a cable with an SFF-8088 connection on one end (for the JBOD end) and whatever the SAS controller has on the other end. I believe you can daisy-chain this setup by connecting a cable appropriately. These parts are available from other suppliers; provantage is linked simply because they have all of them at one store and the price on the case is the best I've found. I haven't received all the parts to try this out yet, but I'm confident that this will work as specified. YMMV. Will [1]: http://www.provantage.com/supermicro-cse-846e1-r900b~7SUPM1YV.htm [2]: http://www.provantage.com/supermicro-cbl-0168l~7SUPA01T.htm [3]: http://www.provantage.com/supermicro-cse-ptjbod-cb1~7SUP9023.htm ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
J4200 is cheap compared to custom made solutions (Supermicro + 3Ware). Ask a Sun reseller for a quote. There is also the Sun Startup Essentials program with great prices on storage systems. Last year I've bought an Infortrend RAID array with custom disks but today I would choose a J4400 + 24 TB. -- 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] j4200 drive carriers
John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. -- richard ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] ZFS extended ACL
On Fri, Jan 30, 2009 at 1:19 PM, Miles Nordin car...@ivy.net wrote: fm == Fredrich Maney fredrichma...@gmail.com writes: fm changing the default toolset (without notification) I wouldn't wish for notification all the time and tell people they cannot move unless they notify everyone, or you will get a bunch of CYA disclaimers and still have no input. Neither would I, nor is that what I asked for. I would however like to see changes like this documented in Release Notes or the Project Page. Unfortunately, that doesn't seem to happen as often as it should. And if you won't take the time to discuss and compromise, you should not be allowed to block people trying to improve things by demanding they seek you out, beg for Your Obstinence's attention, and yeild and appeal for the generosity of your blessing before they can move. Are you looking down your nose at just me, or everyone that is not amused with the it's new, therefor it MUST be better crowd? This is the recipe for a stagnant, bit-rotted system, and it's the *FIRST* attitude the Linux camp threw over the railing. fm from the expected native tools to external tools in order to fm make OpenSolaris more user friendly and familiar to the Linux fm crowd. Yes! I agree it's a very bad reason. I've yelled at not a few Linux people who came to BSD and complained the installer wasn't friendly because it didn't have white-on-blue text. I told them we are building an OS so here are the .tar.gz which is all I ever use, so don't talk to me of this remedial installer. If you want to build installers only, you know where to find the other tent. ``The old tools are unmaintained, ancient and not supporting expected modern standards and features needed to get work done conveniently, and not supporting the stream formats and shell scripts of most other systems (not just Linux ones!), AND don't come with source,'' are several really good reasons. I think you are WAY too easy on yourselves to say these people want something familiar when some of us really want something that works well and is not frozen or abandoned. All of those statements are subjective viewpoints. I look at the feature sets for the ancient tools and see a lot of features that are missing from the GNU alternatives as well as a lot of patches pointing out the lie in unmaintained. Just because they don't have the feature de jour found in the latest unstable build of the so roll your own and smoke it Linux distro doesn't mean that it is unmaintained or ancient. However, just moving the path around is not enough. Whatever tools are going to be the Default ones, the ZFS-specific features need to be added to THOSE tools, not just to some tools somewhere that I have to go hunting for. which is how this thread began. It wasn't about ``OMG a GNUism! why was I not INFORMED?!,'' it was about why is this ZFS feature missing from the Default tool? This thread began because of a feature that was expected, and present in previous versions, was suddenly missing, due to a fairly significant change (in impact, not amount of work involved) that was, to all appearances, arbitrarily made and never communicated. This probably means there should eventually be ONE set of default tools, unless we are really going to meticulously add these ZFS ACL features and every other future architecture to both GNU and ancient usr/bin tools and ancient xpg4 tools and xpg6 tools and ucb tools and ancient 5bin tools, forever. Agreed. Oddly enough, that seems to be the path was taken by Sun quite some time ago with /usr/bin. Those tools are the standard, default tools on Sun systems for a reason: they are the ones that are maintained and updated with new features (some internally developed, some ported from elsewhere where possible and useful). [...] fpsm ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On Sun, Feb 1, 2009 at 2:24 PM, Richard Elling richard.ell...@gmail.comwrote: John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. -- richard I don't think anyone is asking Sun to service disks that aren't theirs, but to claim that's a reason for not selling drive trays is crap. You already claimed in the other thread that Sun has contracts for custom disks so they don't have to worry about short stroking/right-sizing so it should be pretty frigging obvious to support if the disks in the JBOD are Sun or some random crap purchased from fry's. --Tim ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On Sun, 01 Feb 2009 14:45:28 -0600 Tim t...@tcsac.net wrote: On Sun, Feb 1, 2009 at 2:24 PM, Richard Elling richard.ell...@gmail.comwrote: John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. -- richard I don't think anyone is asking Sun to service disks that aren't theirs, but to claim that's a reason for not selling drive trays is crap. No, it's not. To claim that it was the *sole* reason might be, but that's not what Richard said. James -- Senior Kernel Software Engineer, Solaris Sun Microsystems http://blogs.sun.com/jmcp http://www.jmcp.homeunix.com/blog ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Oracle raw volumes
Stefan, one question related with this: would KAIO be supported on such configuration ? Yes, but not as one might expect. As seen from the truss output below, the call to kaio() fails with EBADFD, a direct result of the fact that for ZFS its cb_ops interface for asynchronous read and write are set to nodev, resulting in ENXIO being returned. The kaio() routine detects this ENXIO error, allocates a LWP thread to perform the now synchronous I/O, asynchronously. Although the following following CR does not explicitly call it out, http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6496357 , it resulting in the following changes being made to zfs_ioclt.c and zvol.c, being the removal of the kernel mode AIO interfaces. http://cvs.opensolaris.org/source/diff/onnv/onnv-gate/usr/src/uts/common/fs/zfs/zfs_ioctl.c?r1=%2Fonnv%2Fonnv-gate%2Fusr%2Fsrc%2Futs%2Fcommon%2Ffs%2Fzfs%2Fzfs_ioctl.c%403638r2=%2Fonnv%2Fonnv-gate%2Fusr%2Fsrc%2Futs%2Fcommon%2Ffs%2Fzfs%2Fzfs_ioctl.c%403444 http://cvs.opensolaris.org/source/diff/onnv/onnv-gate/usr/src/uts/common/fs/zfs/zvol.c?r1=%2Fonnv%2Fonnv-gate%2Fusr%2Fsrc%2Futs%2Fcommon%2Ffs%2Fzfs%2Fzvol.c%403638r2=%2Fonnv%2Fonnv-gate%2Fusr%2Fsrc%2Futs%2Fcommon%2Ffs%2Fzfs%2Fzvol.c%403461 These routines were removed as part of 6496357, since both invoked zvol_strategy, a routine the is somewhat synchronous in nature. Even when invoked from multiple LPW threads, parallel threads can still synchronize, depending on byte-range-locking at the following location. http://cvs.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/uts/common/fs/zfs/zvol.c#1130 - Jim At least trying to open /dev/zvol/rdsk/datapool/master where master is defined as: # zpool create -f datapool mirror c1t1d0 c2t0d0 # zfs create -V 1gb datapool/master # ls -lrt /dev/zvol/rdsk/datapool/master lrwxrwxrwx 1 root root 39 Feb 1 18:29 /dev/zvol/rdsk/datapool/master - ../../../../devices/pseudo/z...@0:3c,raw resulted with: # truss -t kaio,lwp_create ./test-kaio /dev/zvol/rdsk/datapool/master kaio(AIOINIT) = 0 lwp_create(0x08046860, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804685C) = 2 /2: lwp_create() (returning as new lwp ...) = 0 /1: kaio(AIOREAD, 5, 0x08046C2C, 1024, 0, 0x0804702C) Err#81 EBADFD /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 3 /3: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 4 /4: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 5 /5: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 6 /6: lwp_create() (returning as new lwp ...) = 0 /1: lwp_create(0x08046820, LWP_DAEMON|LWP_DETACHED|LWP_SUSPENDED, 0x0804681C) = 7 /7: lwp_create() (returning as new lwp ...) = 0 /4: kaio(AIONOTIFY) = 0 /1: kaio(AIOWAIT, 0x, 0) = 1 aio succeeded Comments ? The original question came from a real setup using Sybase which spits a lot of: 733/1: 4137604 6 1 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137611 7 3 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 1 0 733/1: 4137614 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137622 5 2 recv(0x17, 0x100044F20F0, 0x800) = 109 0 733/1: 4137626 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137631 6 2 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137633 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137636 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137637 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137641 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137642 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137646 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137647 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137650 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137652 3 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137655 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 733/1: 4137657 4 0 kaio(0x2, 0x, 0x1) = -1 Err#22 733/1: 4137660 4 1 pollsys(0x110B3A8, 0x1A, 0x10009FE6BF0) = 0 0 Thanks, Stefan test-kaio.c http://sunsite.uakom.sk/sunworldonline/swol-07-1998/swol-07-insidesolaris.html -- This message posted
Re: [zfs-discuss] j4200 drive carriers
On Sun, 1 Feb 2009, Tim wrote: I don't think anyone is asking Sun to service disks that aren't theirs, but to claim that's a reason for not selling drive trays is crap. You already claimed in the other thread that Sun has contracts for custom disks so they don't have to worry about short stroking/right-sizing so it should be pretty frigging obvious to support if the disks in the JBOD are Sun or some random crap purchased from fry's. I agree that this is cause for concern. For example in the past six months Sun has silently dropped almost all of its desktop product line. This includes the Sun Ultra-40 that I bought at great expense. There is no indication that there will be any remaining products in the desktop line and maybe there will be no Sun desktop offering at all once the remaining low-end system is discontinued. Quite a pity since Sun desktops were top quality and price competitive for the features offered. Today I ordered the drive extension bay to support four more drives in this system simply because I don't trust that Sun will provide adequate support for the product it sold to me. So I purchased expansion hardware that I might not even use out of pure unadulterated fear. I have seen how quickly Sun's support for discontinued products vaporizes. I can not afford $6K for the six additional drives at the moment so I can only hope that Sun (or some other vendor) will still be willing to sell SAS drives with appropriate brackets to fit in this system when I am ready to buy them. Bob == Bob Friesenhahn bfrie...@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/ GraphicsMagick Maintainer,http://www.GraphicsMagick.org/ ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On February 1, 2009 12:24:08 PM -0800 Richard Elling richard.ell...@gmail.com wrote John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. i don't agree. disk drives, well non-Sun branded ones, are cheap enough to just keep on hand. even for my home systems, i always buy an extra hard drive of the same size/type the system came with. i expect drives to fail, and hence i used SVM or now for the past few years, zfs, for availability. that covers a failure immediately until such time, usu. 1 day max, that the physical drive can be swapped. a service contract for a hard drive seems a complete waste of dollars, esp. when you consider the 5x markup on the drive. once i've swapped in my cold spare, if i can't wait for a warranty replacement i simply buy another drive. for parts that might be hard to get, or very expensive, a service contract makes great sense. but for hard drives? -frank ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Two-level ZFS
I realize that this configuration is not supported. What's required to make it work? Consider a file server running ZFS that exports a volume with Iscsi. Consider also an application server that imports the LUN with Iscsi and runs a ZFS filesystem on that LUN. All of the redundancy and disk management takes place on the file server, but end-to-end error detection takes place on the application server. This is a reasonable configuration, is it not? When the application server detects a checksum error, what information does it have to return to the file server so that it can correct the error? The file server could then retry the read from its redundant source, which might be a mirror or might be synthentic data from RAID-5. It might also indicate that a disk must be replaced. Must any information accompany each block of data sent to the application server so that the file server can identify the source of the data in the event of an error? Does this additional exchange of information fit into the Iscsi protocol, or does it have to flow out of band somehow? -- -Gary Mills--Unix Support--U of M Academic Computing and Networking- ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On February 1, 2009 12:02:11 PM -0800 Frédéric VANNIERE f.vanni...@planet-work.com wrote: J4200 is cheap compared to custom made solutions (Supermicro + 3Ware). i can't see how that math works. by the time i fully populate a J4200 (12TB), Sun charges in the $10k neighborhood. i can do this using Promise or Xtore or whatever for $3k. Last year I've bought an Infortrend RAID array with custom disks but today I would choose a J4400 + 24 TB again, the Sun enclosure is about the same price as any other enclosure, but the Sun drives make 24TB cost ... well the Sun store is down right now, but I'll estimate $30k or just to be generous let's use their advertised figure of $1/GB, so $24k. An Xtore with 24TB runs $11k. -frank ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Bad sectors arises - discs differ in size - trouble?
You have a raid with 5 terabyte discs. Now some bad sectors arises, so the discs differ in size. What happens with the ZFS raid? Will there be seriuos trouble? Or is this only a problem when ZFS raid is 100% full? Of old, in Linux you didnt allocate the entire disc. Instead you let 100MB be free so there would be headroom for bad sectors. If there are bad sectors the discs will differ in size. Is this a problem? -- 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] Two-level ZFS
On Sun, Feb 01, 2009 at 04:26:13PM -0600, Gary Mills wrote: I realize that this configuration is not supported. What's required It would be silly for ZFS to support zvols as iSCSI LUNs and then say you can put anything but ZFS on them. I'm pretty sure there's no such restriction. (That said, I can't speak for the ZFS team, and it's remotely imaginable -barely- that there could be such a limitation, but it would be strange indeed.) to make it work? Consider a file server running ZFS that exports a volume with Iscsi. Consider also an application server that imports the LUN with Iscsi and runs a ZFS filesystem on that LUN. All of the redundancy and disk management takes place on the file server, but end-to-end error detection takes place on the application server. This is a reasonable configuration, is it not? IMO it's no more and no less reasonable than putting a DB on an iSCSI LUN backed by ZFS. When the application server detects a checksum error, what information does it have to return to the file server so that it can correct the error? The file server could then retry the read from its redundant source, which might be a mirror or might be synthentic data from RAID-5. It might also indicate that a disk must be replaced. ZFS relies on doing all the mirroring and/or RAID-Z work itself in order to be able to recover from bad blocks. If you put the redundancy below ZFS (e.g., using HW RAID) then you typically lose that capability, BUT with ZFS on the target side the target-side ZFS will be able to do the recovery while still retaining end-to-end integrity protection. Neat, eh? Must any information accompany each block of data sent to the application server so that the file server can identify the source of the data in the event of an error? In this case there's no need: the target will know which blocks are bad and automatically correct (assuming enough redundancy exists). Nico -- ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Bad sectors arises - discs differ in size - trouble?
On Sun, Feb 1 at 14:45, Orvar Korvar wrote: You have a raid with 5 terabyte discs. Now some bad sectors arises, so the discs differ in size. What happens with the ZFS raid? Will there be seriuos trouble? Or is this only a problem when ZFS raid is 100% full? Of old, in Linux you didnt allocate the entire disc. Instead you let 100MB be free so there would be headroom for bad sectors. If there are bad sectors the discs will differ in size. Is this a problem? Disk vendors already maintain enough spare capacity so that the drive can transparently provide defect management for a few hundred thousand bad sectors without having to report that it's smaller. If the device reports a smaller capacity than sold originally, short of things like HPA or DCO, it has likely exhausted its spare pool, and it's time to RMA the drive with some urgency. -- Eric D. Mudama edmud...@mail.bounceswoosh.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
Tim wrote: On Sun, Feb 1, 2009 at 2:24 PM, Richard Elling richard.ell...@gmail.com mailto:richard.ell...@gmail.com wrote: John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. -- richard I don't think anyone is asking Sun to service disks that aren't theirs, but to claim that's a reason for not selling drive trays is crap. You already claimed in the other thread that Sun has contracts for custom disks so they don't have to worry about short stroking/right-sizing so it should be pretty frigging obvious to support if the disks in the JBOD are Sun or some random crap purchased from fry's. The drives that Sun sells will come with the correct bracket. Ergo, there is no reason to sell the bracket as a separate item unless the customer wishes to place non-Sun disks in them. That represents a service liability for Sun, so they are not inclined to do so. It is really basic business. -- richard ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On Sun, February 1, 2009 14:24, Richard Elling wrote: John-Paul Drawneek wrote: the J series is far to new to be hitting ebay yet. Any alot of people will not be buying the J series for obvious reasons The obvious reason is that Sun cannot service random disk drives you buy from Fry's (or elsewhere). People who value data tend to value service contracts for disk drives. The people who in fact want service contracts covering their drives will of course buy from Sun; that works well for them, and well for Sun. Win win! Thing is, the people who *don't* want service contracts covering their drives (which, in my experience, is everybody in the SATA era; it's a silly waste of money) will be pissed off, and will have to avoid buying that Sun product. This is a lose for Sun, and possibly a lose for them. Will anybody actually be forced to buy their drives from Sun? If so, is that something for Sun to be proud of? It looks to me as if this kind of policy doesn't bring business to Sun, it drives it away. At least, it doesn't bring *happy* business to Sun; it might bring people feeling coerced and looking for an excuse to escape. Google, who live and die by their data, prefer to use cheap commodity hardware. Since they're *already* set up with redundancy, they can afford a higher failure rate. This sort of thinking is true of a lot of high-end places these days, their reliability requirements are so high that one good server (or disk drive) can't possibly meet them anyway. -- David Dyer-Bennet, d...@dd-b.net; http://dd-b.net/ Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/ Photos: http://dd-b.net/photography/gallery/ Dragaera: http://dragaera.info ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On Sun, 1 Feb 2009, Richard Elling wrote: The drives that Sun sells will come with the correct bracket. Ergo, there is no reason to sell the bracket as a separate item unless the customer wishes to place non-Sun disks in them. That represents a service liability for Sun, so they are not inclined to do so. It is really basic business. -- richard This thread has been running for a little too long, considering the issues are pretty simple. Sun sells a JBOD storage product, along with disks and accessories. The disks they provide are mounted in the correct carriers for the array. The pricing of the disks and accessories are part of the price calculation for the entire system - you could provide the array empty, with a full set of empty carriers but the price will go up. No brand name storage vendor supports or encourages installation of third party disks. It's not the way the business works. If the customer wants the reassurance, quality, (etc, etc) associated with buying brand name storage, they purchase the disks from the same vendor. If price is more critical than these factors, there's a wide range of white box solutions on the market. Try approaching IBM, HP, EMC, HDS, NetApp or similar and ask to buy an empty JBOD and spare trays - it's not happening. Yes, this is unfortunate for those who would like to purchase a Sun JBOD for home or for a microbusiness. However, these users are probably aware that if they want to buy their own spindles and run an unsupported configuration, their local metal shop will be happy to bang out some frames. Not to mention the fact that one could always run up a set of sleds in the shed without too much strife - in fact, in the past when spuds were less commonly available, I've seen a home user make sleds out of wood that did the job. Now, I'm looking forward to seeing the first Sun JBOD loaded up with CNC-milled mahogany sleds. It'll look great. -- Andre van Eyssen. mail: an...@purplecow.org jabber: an...@interact.purplecow.org purplecow.org: UNIX for the masses http://www2.purplecow.org purplecow.org: PCOWpix http://pix.purplecow.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
+-- | On 2009-02-01 16:29:59, Richard Elling wrote: | | The drives that Sun sells will come with the correct bracket. | Ergo, there is no reason to sell the bracket as a separate | item unless the customer wishes to place non-Sun disks in | them. That represents a service liability for Sun, so they are | not inclined to do so. It is really basic business. A specific example of why this policy drives me crazy: X4150. 8 disk bays. SAS HBA, SATA onboard. In its initial incarnation, the SAS HBA was optional (this was fixed soon after its release, but I got bitten by it, ordering it before support even knew it was released). I had also read that Sun would be selling 2.5 SATA disks for the system by the end of the year (2008). Regardless, mixing SAS and SATA in this system would would be incredibly useful for any number of applications. Sun does not sell 2.5 SATA disks for the X4150, and I cannot purchase sleds. Further: X4150s are presumably commonly deployed for database work (that's what both of mine do, along with a lot else). According to zilstat (thank you, Richard!), r/w SSDs would prove beneficial. Sun does not sell SSDs for the X4150, and I cannot purchase sleds. Obviously, I could buy 73GB SAS disks and simply repurpose their sleds. I hope everyone (who is a consumer) can agree that sucks. rant The company I work for is a small ESP; we've been around forever (1995), and until I moved us to Solaris two years ago, we were a pure Linux shop. We like Sun software. We like Sun hardware. We had a (small) part to play in getting DTrace into Perl 5.10. We're proponents of the company and its technology. As with everyone else who has asked for this sort of thing, I just wish Sun would offer me a bit more flexibility so I can more effectively do my job using their gear. (I went the eBay route for my X4100 sleds; the way the numbers came out in the end, it didn't really make a lot of sense to do it again, so I haven't. But I can actually buy the disks I wanted for my X4100s from Sun.) Like all small fish, we have to move fast and be creative with our purchasing and infrastructure. Sun's focus on large deployments and Corporate can often leave us little guys in the cold. Given Sun's business focus and climate (as Richard says, it's basic business), I've mostly learned to suck it up, but that doesn't make things like this taste any less bitter. /rant -- bda Cyberpunk is dead. Long live cyberpunk. http://mirrorshades.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Two-level ZFS
Gary, I realize that this configuration is not supported. The configuration is supported, but not in the manner mentioned below. If there are two (or more) instances of ZFS in the end-to-end data path, each instance is responsible for its own redundancy and error recovery. There is no in-band communication between one instance of ZFS and another instances of ZFS located elsewhere in the same end-to- end data path. A key understanding is the a ZVOL provides the same block I/O semantics as any other Solaris block device, therefore when a ZVOL is configured as an iSCSI Target, and the target is accessed by an iSCSI Initiator LU, there is no awareness that a ZVOL is the backing-store of this LU. Although not quite the same , this ZFS discussion list raises questions about configuring ZFS on RAID enable storage arrays, and how using simple JBODs might be a better solution. Jim Dunham Engineering Manager Sun Microsystems, Inc. Storage Platform Software Group What's required to make it work? Consider a file server running ZFS that exports a volume with Iscsi. Consider also an application server that imports the LUN with Iscsi and runs a ZFS filesystem on that LUN. All of the redundancy and disk management takes place on the file server, but end-to-end error detection takes place on the application server. This is a reasonable configuration, is it not? When the application server detects a checksum error, what information does it have to return to the file server so that it can correct the error? The file server could then retry the read from its redundant source, which might be a mirror or might be synthentic data from RAID-5. It might also indicate that a disk must be replaced. Must any information accompany each block of data sent to the application server so that the file server can identify the source of the data in the event of an error? Does this additional exchange of information fit into the Iscsi protocol, or does it have to flow out of band somehow? -- -Gary Mills--Unix Support--U of M Academic Computing and Networking- ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
On Sun, 1 Feb 2009, Bob Friesenhahn wrote: I am worried that Sun is primarily interested in new business and tends to not offer replacement/new drives for as long as the actual service-life of the array. What is the poor customer to do when Sun is no longer willing to offer a service contract and Sun is no longer willing to sell drives (or even the carriers) for the array? You can still procure replacement drives for real vintage kit, like the A1000/D1000 arrays. I doubt your argument is valid. As a side point, by the time these arrays are dead buried, the sleds for them will no doubt be as common as spuds (and don't we all have at least 30 of those lying around?) Sometimes it is only a matter of weeks before Sun stops offering supportive components. For example, my Ultra 40 was only discontinued a month or so ago but already Sun somehow no longer lists memory for it (huh?). If your trusty Sun partner couldn't supply you with memory for an Ultra-40, I'd take that as a sign to find a new partner. EOSL products vanish from websites but parts can still be ordered for them. -- Andre van Eyssen. mail: an...@purplecow.org jabber: an...@interact.purplecow.org purplecow.org: UNIX for the masses http://www2.purplecow.org purplecow.org: PCOWpix http://pix.purplecow.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
Bryan Allen wrote: +-- | On 2009-02-01 16:29:59, Richard Elling wrote: | | The drives that Sun sells will come with the correct bracket. | Ergo, there is no reason to sell the bracket as a separate | item unless the customer wishes to place non-Sun disks in | them. That represents a service liability for Sun, so they are | not inclined to do so. It is really basic business. A specific example of why this policy drives me crazy: X4150. 8 disk bays. SAS HBA, SATA onboard. In its initial incarnation, the SAS HBA was optional (this was fixed soon after its release, but I got bitten by it, ordering it before support even knew it was released). I had also read that Sun would be selling 2.5 SATA disks for the system by the end of the year (2008). Regardless, mixing SAS and SATA in this system would would be incredibly useful for any number of applications. Sun does not sell 2.5 SATA disks for the X4150, and I cannot purchase sleds. Further: X4150s are presumably commonly deployed for database work (that's what both of mine do, along with a lot else). According to zilstat (thank you, Richard!), r/w SSDs would prove beneficial. Sun does not sell SSDs for the X4150, and I cannot purchase sleds. Obviously, I could buy 73GB SAS disks and simply repurpose their sleds. I hope everyone (who is a consumer) can agree that sucks. rant The company I work for is a small ESP; we've been around forever (1995), and until I moved us to Solaris two years ago, we were a pure Linux shop. We like Sun software. We like Sun hardware. We had a (small) part to play in getting DTrace into Perl 5.10. We're proponents of the company and its technology. As with everyone else who has asked for this sort of thing, I just wish Sun would offer me a bit more flexibility so I can more effectively do my job using their gear. +1 The astute observer will note that the bracket for the X41xx family works elsewhere. For example, http://sunsolve.sun.com/handbook_pub/validateUser.do?target=Systems/SunFireX4150/components#Disks XRA-SS2CF-73G10K contains 541-2123 http://sunsolve.sun.com/handbook_pub/validateUser.do?target=Systems/7410/components#Disks XTA7410-LOGZ18GB contains 570-1182 XTA7410-READZ100G contains 541-2123 But that is just the mechanical parts. Unfortunately, there is also firmware to be considered, though that tends to solve itself over time. One would think that Sun could have leveraged the common chassis, but they do not seem capable of doing so. Don't look for logic here, there is none. -- richard ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] j4200 drive carriers
+-- | On 2009-02-01 20:55:46, Richard Elling wrote: | | The astute observer will note that the bracket for the X41xx family | works elsewhere. For example, | http://sunsolve.sun.com/handbook_pub/validateUser.do?target=Systems/SunFireX4150/components#Disks | XRA-SS2CF-73G10K contains 541-2123 | http://sunsolve.sun.com/handbook_pub/validateUser.do?target=Systems/7410/components#Disks | XTA7410-LOGZ18GB contains 570-1182 | XTA7410-READZ100G contains 541-2123 I had noted that, when looking for the sleds alone, though not those specific cases. Useful! | But that is just the mechanical parts. Unfortunately, there is also | firmware to be considered, though that tends to solve itself over time. Hopefully! | One would think that Sun could have leveraged the common chassis, | but they do not seem capable of doing so. Don't look for logic here, | there is none. mm. -- bda Cyberpunk is dead. Long live cyberpunk. http://mirrorshades.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss