Re: performance limitations of linux raid
bug1 wrote: > > > > > I don't believe the specs either, because they are for the "ideal" case. > > However, I think that either your benchmark is flawed, or you've got a > > crappy controller. I have a (I think) 5400 RPM 4.5GB IBM SCA SCSI drive in > > a machine at home, and I can easily read at 7MB/sec from it under Solaris. > > Linux is slower, but that's because of the drivers for the SCSI controller. > > I haven't done any benchmarks on my IDE drives because I already know that > > they're SLOW. > > Greg > > > > Whatever you think of the interface (ide vs scsi) you have to accept > that a drives speed is dependent on its rotation speed. Not entirely true. RPMs is but _one_ factor, not the determining factor. > > A 7200RPM IDE drive is faster than a 5400RPM SCSI drive and a 1RPM > SCSI drive is faster than a 7200RPM drive. Not always true, if you are talking about data throughput and access speed. > > If you have two 7200RPM drives, one scsi and one ide, each on there own > channel, then they should be about the same speed. > > Multiple drives per channel give SCSI an edge purely because thats what > the scsi bus was designed for. You pay a big dollars for this advantage > though. As well as various _other_ advantages, see otehr post. -- In flying I have learned that carelessness and overconfidence are usually far more dangerous than deliberately accepted risks. -- Wilbur Wright in a letter to his father, September 1900
Re: performance limitations of linux raid
Seth Vidal wrote: > > > A 7200RPM IDE drive is faster than a 5400RPM SCSI drive and a 1RPM > > SCSI drive is faster than a 7200RPM drive. > > > > If you have two 7200RPM drives, one scsi and one ide, each on there own > > channel, then they should be about the same speed. > > > > Not entirely true - the DMA capabilities of IDE could provide faster > transfer modes than your avg scsi card could generate. > > I have a 7200 RPM LVD scsi drive and a 7200RPM UDMA ide drive and the IDE > wins EVERY SINGLE TIME. > > -sv I think everyone seems to be missing the _point_ of SCSI. SCSI is NOT just about raw speed, even though some SCSI has a speed of 160MB/second. SCSI has many advantages over IDE, oe EIDE, such as command queing. Let's see your IDE drive handle 10 or 15 *simultaneous* reads/writes. let's see it have a MTBF of over 1 Million hours. Let's see it connected along with 13 other drives on the same chain. Let's see it run on cables running for several yards or meters. Let's see it conected to multiple computers. Heck, just for fun, let's run IP over IDE. Oh, wait, that's SCSI that can do that ;) The point is, comparing speed of SCSI vs any IDE variant is like comparing apples and oranges. That said, copmparing two drives of any variant, and basing their performance upon the rotational speed is also an error. RPMs are not the sole determining factor. Other factors include the size of the drive in MB/GB, and the seek or access time. I have, for example, a 5400RPM SCSI2 drive that still outperforms 7200RPM IDE drives colleagues have. I have seen 6X SCSI CDROMs that perform better than 8X EIDEs. I have a 7200RPM SCSI outperforming 10Krpm IDE drives. The determining factor in the choice of SCSI vs. IDE should be in what the machine will be doing. If all you need is a desktop machine fo rgaming and/or basic sufing/office type work, you probably don't need SCSI. if you are going to run a server that may be running some intensive I/O, and need reliability and longevity, you should be looking at SCSI, not IDE. After all, is speed is your sole criteria, chances are you are missing something. ;^) -- In flying I have learned that carelessness and overconfidence are usually far more dangerous than deliberately accepted risks. -- Wilbur Wright in a letter to his father, September 1900
Re: New-style raid under 2.3.X?
Dave Wreski wrote: > > > Just get the Alan Cox patches from your favorite kernel.org mirror, a > > clean 2.2.13 kernel tree, and apply. No failures, just good clean > > raiding. > > Is it in fact the case that the pre-patch-2.2.14-11 (released today) patch > has the latest raid drivers in it? That sure seems to be what you are > stating, but I recall there being some issues with it requiring newer > user-level programs that was causing a delay in merging with the latest > kernel? No, I was referring to linux-2.2.13ac[1-3], which does include the good raid. -- In flying I have learned that carelessness and overconfidence are usually far more dangerous than deliberately accepted risks. -- Wilbur Wright in a letter to his father, September 1900
Re: New-style raid under 2.3.X?
Bill Anderson wrote: > > Anders Qvist wrote: > > > > I would prefer to compile a 2.3.13+ kernel in order to get the most out > > of my Matrox G400 (dual) graphics card. However, I have a system booting > > off a raid 0 partition, so I need the new-style raid. There does not seem > > to be raid0145 patches for 2.3 kernels and patching the 19990824 version > > onto 2.3.13+ kernels has a few failiures too many to easily fix. What do I > > do? > > Just get the Alan Cox patches from your favorite kernel.org mirror, a > clean 2.2.13 kernel tree, and apply. No failures, just good clean > raiding. Ooops, I saw 2.3.13, but the brain read 2.2.13 ... sorry. -- In flying I have learned that carelessness and overconfidence are usually far more dangerous than deliberately accepted risks. -- Wilbur Wright in a letter to his father, September 1900
Re: New-style raid under 2.3.X?
Anders Qvist wrote: > > I would prefer to compile a 2.3.13+ kernel in order to get the most out > of my Matrox G400 (dual) graphics card. However, I have a system booting > off a raid 0 partition, so I need the new-style raid. There does not seem > to be raid0145 patches for 2.3 kernels and patching the 19990824 version > onto 2.3.13+ kernels has a few failiures too many to easily fix. What do I > do? Just get the Alan Cox patches from your favorite kernel.org mirror, a clean 2.2.13 kernel tree, and apply. No failures, just good clean raiding. Bill -- In flying I have learned that carelessness and overconfidence are usually far more dangerous than deliberately accepted risks. -- Wilbur Wright in a letter to his father, September 1900
Re: my sunday afternoon nightmare (raid crash)
Niklas Paulsson wrote: > > Thomas Waldmann wrote: > > There is no problem with patching 2.2.11 or 2.2.12 or 2.2.13 with the 0824 > > patches - I (and many other people) verified the rejects and there is no real > > problem - the stuff rejected simply already IS in the standard kernel. > > > > This is also the reason why there is no new patch. > > I'm a solaris admin today, our server park today is 85% solaris 5% linux > and 10% nt. I would just love to switch over to 100% linux. But frankly, > I don't have the guts. > > The missmatch in patch revisions for raidtools is one of the reasons for > that. > > Even if the final result was 100% identical it gives me the creep to run > a production box on home patched kernel, where the kernel number is a > mismatch and I get rejects when aplying it. > > This is probably just me being conservative and boring. But nevertheless > that is the way I feel. So use the Alan Cox patches. :-) Work great for me. One patch, no rejects, lots of goodies. > It would boost my confidence in the raidtools package by a factor of 100 > or so if I knew that it was so well maintained that I always could find > a nice new patch for every new kernel. > > A clear upgrade path would be of some assistance, but I could figure > that one out myself ;) If one lives on the leading edge of kernel releases (.odd or .even), one is expected to be able to deal with patching and applying old patches to new source. The reason that the current 'alpha' code is not in the kernel is totally unrelated to stability, rather to perceived backwards compatibility. A patch section that is rejected due to that change already being incorporated is nothing to worry about. > I have been "evaluting" the raidtools package for our site a couple of > months now on a smaller server (90 Gb of data, kind of a NAS concept). > And what has been the major problem for us have not been the raidtools, > it's the nfs server that has been behaving somewhat strange at times. > > The lack of a journaling fs is also a major show stopper at our site, > once we have a relaible journaling fs that works well together with sw > (and hw) raid , we will begin thinking of a migration path to linux on > our big boxes. But that belongs on another list I am trying out ext3 w/journaling (also alpha) as well as reiserfs w/journaling (alpha) this next few weeks. So far, looks great. :-) Bill -- "They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown." -- Carl Sagan
2.3.1x and raid?
Are there any patches avaiable for this yet? I really want to try out 2.3 :') -- "They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown." -- Carl Sagan
Re: redhat 6.1 RAID: what's different - installs
[EMAIL PROTECTED] wrote: > > On Tue, 5 Oct 1999, Alvin Oga wrote: > > > I'd be interested in writing an installer/duplicator... > > > > - I like to do something like: > > a. give um a bootable cdrom... ( self detect the hardwara ) > > ( aka rescue cdrom ) > > Try the PXE stuff in redhat 6.1. (For some reason, I'm just fascinated by > netbooting; even if 2gig IDE "system" drives cost nothing. > > > > > - the installer should automatically detect the hardware installed... > > ( don't ask about hardware... go and look for it ) > > Go look at 6.1. Its pretty slick. (Yes, it detects hardware.) > > > c. ask about raid setup and backup setup... > > - raid0/raid1, raid5... > > Go look at 6.1. Hmm ... didn't see this in the install yesterday or today ... perhaps I am missing something? -- "They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown." -- Carl Sagan
Re: RAID1 over RAID0 on 2.2.10
Alan Meadows wrote: > > Hello, > > From past messages I've gotten the feeling that some people consider > 2.2.10 unstable just as 2.2.9, with the corrupt filesystem issue. Does > anyone here have experience with 2.2.10 enough to know its stable, or > is anyone firm on the idea that its unstable? None of the previous > e-mails seem to be conclusive enough for me =) I'll add my name to the list of people running 2.2.10 and RAID without problems. Actually, I am using the 2.2.6 patch. I know there is a 2.2.10 patch, but since my setup works, I leave it alone. At least until a new kernel comes out that squashes the TLB IPI Wait Queue bug ... -- Bill Anderson
Re: One or several raid-0 devices on SMP machine?
Helge Hafting wrote: > > I am building a dual processor machine with 2 u2w drives. > It will have a couple of root partitions (one for rescue > purposes) a swap partition on each drive, and the rest > will be raid-0 for performance. > > Is there any performance difference between one big RAID-0 > partition with everyting (/usr, /home,...) or > having several raid-0 devices on these 2 drives? > (Such as having /usr on sda1+sdb1, /home on sda2+sdb2,...) > > I don't need advice on partitioning, only the raid performance. Well, not too sure about the performance side, but on the paranoid side, I use the multiple devices method. That way, if something goes wrong, I don't _neccessarily_ lose everything in one fell swoop. YMMV. __ Bill Anderson
Re: root raid1 support w/o initrd
Thomas Seidel wrote: > > If you are interested in a very simple root raid1 support for kernel 2.2.x and > raidtools 0.42 without the need of initrd, take a look at > ftp://ftp.ddb.de/pub/linux/root_raid1_support/ > You will find there a small patch for drivers/block/md.c to have raid1 setup > for the root partition at boot time and a raid1 stop for the root partition > at the end of a system shutdown. > > I didn't test it, but the patch could also work for raid5 devices. If you like > to check this out, remove the comment brackets "not supported yet" in line 1307 > of the patched md.c and change the md= parameters. > > Thomas I would like mostly to see this as an option in the install process (RH, for example), but alas, don't yet have the time :-(
Re: raid 0.90 on 2.0.37 ?
Giulio Botto wrote: > > On Wed, Jul 07, 1999 at 09:12:02AM +0200, paolo furieri thusly shaped the electrons: > > I am using raid-1 0.90 on 2.2.5-22 redhat kernel, and I experienced a lot of > > instability problem: services (http, syslog, named ...) stay up for some > > hours, then they suddenly go down. Then I am waiting for a new raid patch for > > new kernels. But I have seen in mailing-list that 2.2 kernel has filesystem > > stability problem. > > Hmm, I'm missing these announces, but I've got a RedHat 6.0 2.2.5-22 with > a 3-way SCSI raid5 with an uptime of about a couple of months, and It's hosting > http, ftp, nfs, samba and other services with no problem ... are you sure the > problem is the kernel ? I believe he is referring to some fs instability that has happened in 2.2.10, not 2.2.x in general. Bill
Re: RAID under 2.2.10
Fred Reimer wrote: > > On Tue, 06 Jul 1999, Christoph Martin wrote: > > You can apply the 2.2.6 patches to 2.2.10. But it is not working > > correctly. Normal operation is ok, but if a raid comes out of sync and > > need a resync (like when you reboot without a proper shutdown), this > > would fail. > > > > I have a 2.2.10 with 2.2.6 raid patches running at the moment, because > > I need 2.2.10 to get Informix IDS running. But when the machine > > crashed I had to boot my old 2.2.6 to get the rebuild done and then > > reboot with 2.2.10 for Informix. > > > > Ingo Molnar talked about working on the patches for 2.2.10 and that he > > fixed the problem with resync, but he did not yet release the code. > > > > Christoph > > So > > If you're only using RAID0, which you can not rebuild and has no > syncing, then the 2.2.6 patches applied to 2.2.10 provide tested, > production quality RAID? > > I will take a backup before proceeding, but I would also like to get > concensus from linux-raid participants and particularly Ingo... Well, if it helps, I've been running RAID0 with said patch on 2.2.10 for quite a while now. (knock on wood). Even when having to powercycle the machine without a proper shutdown.
Re: raid on raid
Lawrence Dickson wrote: > > We want to use Linux raid to make a raid out of pre-existing raids, which I > thought was possible. We tried creating two RAID-1s of 2 disks each, then a > RAID-0 of these two joined. We finished the mkraids on the RAID-1s before > starting the RAID-0. It said something like > As mentioned by Brian, this setup is not likey a valid one. Try reversing it: make two raid0 and run them into a raid1. that should work. Bill
Re: Problems setting up raid-0
Kelvin Edwards wrote: > > I have two SCSI disks that I am trying to combine using software raid > disk striping (raid-0). mkraid returns the folowing error: > > /sbin/mkraid --really-force /dev/md0 > DESTROYING the contents of /dev/md0 in 5 seconds, Ctrl-C if unsure! > handling MD device /dev/md0 > analyzing super-block > disk 0: /dev/sdb1, 17782768kB, raid superblock at 17782656kB > disk 1: /dev/sdc1, 17782768kB, raid superblock at 17782656kB > mkraid: aborted > > At this point running raidstart -a generates an Invalid argument > error. Here is the relevant /etc/raidtab: > > raiddev /dev/md0 > raid-level 0 > persistent-superblock 0 > chunk-size 16 > nr-raid-disks 2 > device /dev/sdb1 > raid-disk 0 > device /dev/sdc1 > raid-disk 1 > > Any suggenstions. I am running RedHat 6.0, and kernel 2.2.6 or > kernel 2.2.10 (I tried both no difference). 1) Did you patch the kenrel before you compiled (2.2.6/2.2.10)? 2) Turn on persistent-superblock (chang 0 -> 1) Bill Anderson > > --kelvin
Re: RAID 1 and RH 6.0
Till Mommsen wrote: > > > > 3. Why is RAID so hard to set up on Linux? > > > > Huh? It was a snap for me. Perhaps it's more of an familiarity > > issue. It > > doesn't take me 3 days to set up an FTP server or 2 days for a > > nameserver... > > maybe hours but not days. > > > > That should be a real consideration. Setting up a server that you are not > > sure of and not comfortable with will almost certainly result in a Bad Day > > if something goes wrong. I certainly would not know the first thing about > > setting up NT server with Raid and therefore wouldn't recommend a > > production > > server on it - especially if I would be responsible for it - even > > though it > > may do just fine. > > Well, I don't want to start a basic discussion about Linux pro's and con's, > but I agree with Shon and there are a few things about Linux and RAID that I > feel should should be said. I have started to use Linux in production > environments half a year ago. Now I got servers running root RAID1, Samba, > Amanda, Firewall/dial-up router etc. Nevertheless, I can calculate if > something takes me two hours with NT it will take at least two days with > Linux to really get it running, understand what is going on and to debug it. > Of course, after I set up things with Linux, I can forget the machine, it > runs just fine. And this is an important point fo rme. I would rather spend extra time in setup, than repeatedly spend time 'fixing' things. > > I really like Linux (although it somethimes occurs to me that Linux maybe > doesn't like me...), but at this point, I expect a big "bang" to Linux: It > has been praised by the press. Lots of Win-Admins will suggest Linux to > their IT Managers. But what will they hear, if, after maybe three weeks of > experimenting (and that's the minimum it will take for a Win-Linux > converter), there is still nothing really presentable? Will the positive > press messages continue? What, if the positive publicity turns, because > Linux is stamped "command line gurus only"? No different, since we started, and still have, that stamp. > > Marc, you are saying you wouldn't want to setup a RAID1 on NT. Have you > tried it? Since you know computers I'd think it will take you 30 minutes the > first time. Just because that's the time you need to find the right program > in the help files. Not beacause you will encounter any trouble. After the > first time, you'll do it in five minutes while you are having a pizza (if > you can eat that fast). A beginner will need two hours. RAID handling is not > much more difficult. Aside from the time it toook to actually compile the kernel, raid took me all of about 5 minutes on Linux (counting the kernel, still under ten :). This is one thing no UI, whether graphical or command line, can anticipate well. Many of the things that take mere seconds or minutes in NT for me take significantly longer in NT. (OT side note: yesterday I tried navigating and using the file browser in GNOME, and was having difficulty remembering where things were at on the fs. ...). Bet you could set up S-RAID in Linux in 5 minutes now too ;-) > > I really do not want to critize anybody here (and I certainly wouldn't feel > entiteled to). But, after following this list (and others) for six months > now, I've seen 11 Kernels and thousands of traps, hints and methods how to > setup RAID. Redhat doesn't seem to be able to include an easy and correct > RAID setup in their distribution. SuSE and others don't even mention > anything about it. Why? But, what's one of the first thing somebody does, > who sets up a Server? - He thinks about RAID. It's nice that SW RAID is > faster and - theoretically - cheaper than HW. But what, if you calculate the > time it takes you to implement SW RAID? The lack of a distibution putting out a properly configured distribution is not really a reflection on the raidtools. RH tried for bragging rights, I imagine, and because they are aiming at corporate entry. For me, the first thing I think about when setting up a server is: security. But, ask 100 admins ...;^) But will you do it more than once? How often will you set up a software raid array? If you do it frequently, the time/cost to learn it amortizes well. Additionally, you would factor in the cost of learning how to get NT to cope with the particular hardware choice. How much would this run at ~400$US/call to MS? (not always neccesary, but frequently). > > This is not a speech for GUI's and MS. I started Linux because I didn't like > Bill and the way MS is running things. I was sick of protection faults, user > profile spying, monopol, and so on. And so far, I struggeld my way through. > Thanks to your help here. I am sure, others with the same profile have and > will follow. But: how many "newbie questions" can you support? As many as need be. :-) Remembering that raidtools is still < 1.0, many of these issues are still developing. As the tools mature, other aspects, such as a gu
Re: Raid problems.
Chris Price wrote: > > Robert, why are you running raid on 1 disk??? > > What benefit do you expect to derive from running raid on a single > disk? > > Unless you have a special application, there is **NO** point to > creating a raid array from one disk. > > Chris > Actually, I can think of a few. o Learning Granted, no speed improvements, but you can learn about it. Knowledge like that comes in handy in Interviews ;-) o Partitioning issues. It doesn't appear to be the case in this situation, but if you have a large disk, with multiple partitions (for some reason), you could use it to have a single directory. Granted there are other means, and I wouldn't reccomend doing it unless it was the more expedient method at the tim. o Testing? This one kind of goes aling with learning. Perhaps you are considering, or testing it out to see if it works with new kernels, etc.. Just off the top of my head ... :-) Bill
