On Fri, 25 Nov 2005 18:06:44 -0500, Alex Kirk <[EMAIL PROTECTED]> wrote:
>Hello All, > >I just picked up three Maxtor 200GB UltraATA/133 drives that I'd like to use in >a RAID-5 configuration with my OpenBSD web/mail server. I've spent the past >couple of hours carefully combing through all of the RAID cards that are listed >as supported on http://www.openbsd.org/i386.html#hardware, and unless I'm >missing something, there's something of a gap between ATA/100 and SATA/150 >there >(for example, the MegaRAID 150-4/6 and the MegaRAID i4). I've even seen from >the >archives that the MegaRAID 133-2 is *not* supported, due to its lack of an >integrated I/O processor. That said, I apologize if this question has already >been answered, but what would be my best bet for building a RAID-5 with these >disks under OpenBSD? > >Thanks, >Alex Kirk Well, late as usual, I'm finally getting around to rolling out 3.8 from the Official OpenBSD Release CD, and oddly enough, I was just testing out a "MegaRAID ATA 133-2" on one of the boxes. You are partially correct about the "MegaRAID ATA 133-2" HBA; the "MegaRAID ATA 133-2" is _partially_ supported in OpenBSD mainly because it is only _partially_ a RAID device. The low-end ATA-133-2 card is actually a "fake-RAID" device that requires specialized drivers to enable the RAID functionality. Though the "ATA 133-2" card can not do it's usual "fake-RAID" on OpenBSD due to the lack of "fake-RAID" drivers, the card still works perfectly as standard ATA-133 controller. -Saying it is "*not*" supported is somewhat unfair, since it actually is supported as well as it can be without adding a closed source binary driver to your system. If you really needed "RAID" functionality and it was the only card you had, you could easily use it as a normal controller then do SoftRAID on top of it. Personally, I've liked the "MegaRAID i4" and rolled them out in 3.6/3.7 systems. They work really well and the suggestion to use them came from someone on this list (C. Bernsend I think). Don't fool yourself into thinking the ATA100 versus ATA133 thing is actually some big deal. It's not. Sustained transfer rates of a single drive are generally between 25 and 40 MB/s, so even having two drives on a single channel does not exceed the max transfer rate of the channel. -You've been caught up by the MAXTOR marketing ploy. Similar nonsense is even more true for the SATA150/SATA300 marketing crap; the sustained transfer rate of a single drive can not saturate the max transfer rate of the interface. In contrast to past experience, at the moment I'm currently fighting with a new SuperMicro dual 1Ghz PIII system that is refusing to boot due to the "MegaRAID i4" card. There is something goofy going on with it. MainBoard: SuperMicro P6DGE Procs: (2) P-III 1Ghz RAM: 2GB RAID_HBA: LSI Logic MegaRAID i4 Firmware: Current (version "N661") Disks: (4) MAXTOR 6Y250P0 250GB EIDE Physical: One Disk Per EIDE Channel Logical: 1 Logical Device (RAID 5 across all four disks) Checks: Consistency Check Currently Running... BootMedia: Official OpenBSD 3.8 Release CD-ROM DMESG: (still waiting for CC) PROBLEM: Seems to be waiting in an endless loop ami0: timeout ccb ami0: timeout ccb ami0: timeout ccb ... I'll post a DMESG as soon as I can, probably in a new thread. It will be a while since I'm waiting the 20+ hour long "Consistency Check" to complete before hooking up serial to capture the DMESG. If you ever hit a problem, always check everything you can before wasting developer time with bug report that may be caused by _you_ or _your_ faulty hardware. Also, it's very unfair to report a bug unless you are willing to do the work necessary to help the developers replicate the problem and do the needed testing for them. If you search the list archives and CVS logs you'll see that the problem I am having has been worked on by both Theo@ and [EMAIL PROTECTED] One of the people who posted a "ami0: timeout ccb" problem on a "LSI MegaRAID STAT 150-4" HBA in May to this list was unwilling to actually help Marco test and debug the issue. Most of my personal hardware is ancient by modern standards (<= 1GHz and I kinda like it that way), so if you go with a MegaRAID i4 you probably won't have the same issue I'm currently having, or it might be fixed by the time you actually get the card. Kind Regards, JCR
