i have 8 smp machines (dual p3 450mhz, 512M ram) using intel mb with onboard
ncr876 scsi, and two 9gig quantums in each. running software raid1 on all of
the disk, except small swap and /boot partition on each disk. using kernel
2.2.12 with the 2.2.11 alpha raid patch.
also have a couple of older
there are know side effects of running vmware with pretty much anything.
you cannot trust binary only kernel modules, especially if you are not using
'exactly' the same compilation of the kernel that they did for their testing.
(if they did any testing, how do you know without the code?)
are you
john b said:
> Performance is pretty good - these numbers are for a first generation
> smartcan (spring '99)
these numbers are also useless since they are much too close to your ram size,
and bonnie only shows how fast your system runs bonnie :) a better benchmark
would be to see how this runs w
the linux kernel stripes swap writes across multiple swap partitions of the
same priority. striping (effectively raid0) provides NO reliability in case of
disk failure. this is what the original poster was asking about.
to aswer the original question, i have twice tried to make swap on raid1,
whi
Kelly French <[EMAIL PROTECTED]> said:
> For me, I'm just tarring up hosts to tape. When the tape develops an
> error, and I use software compression (i.e. -zvvcf /dev/nst0), then I'll
> lose the whole session. If I let the tape drive compress, it won't
> compress as good (like you said), but
Wondering if anyone out there has any experience with the Acceleraid 150?
this is included in some va-linux/research servers, with 16 megs of ram.
i am trying to determine what the support for these is like under linux
(drivers aside- how about the setup software, etc), as well as how they
perfo
> I\'m looking to create a small (3-disk) raid-5 array. Each disk
will
> probably
> be a 9 or 18gb drive. This is for a small news/messaging server.
I\'m
> looking
> for a cost/performance solution that will give me high random I/O.
have you considered raid1? i wonder if that might give fast
Giulio Botto <[EMAIL PROTECTED]> said:
> On Fri, Apr 30, 1999 at 10:44:59PM -0400, m. allan noah thusly
shaped the electrons:
> > try
> > raidhotadd /dev/md0 /dev/hdjx
> > raithotadd /dev/md0 /dev/hdix
>
> I\'m afraid this will not work as the md device is not started yet
...
was this abou
Thornton Prime <[EMAIL PROTECTED]> said:
> I installed software RAID on a RedHat 5.2 machine last week. I had
> previously upgraded the kernel to 2.2.3. Everything went well.
>
i take it you did not patch the kernel before you installed it?
> It is RAID5 on 5 Seagate SCSI drives on a BusLogic
yeah- get scsi and throw that ide crap away :)
actually, i have found that the ability to recover from ide disconnect depends
on the bios and the drive. i found this to work with some seagate drives a
year or two ago, as long as the bios was award. would not work with ami bios
or any other brand
Osma Ahvenlampi <[EMAIL PROTECTED]> said:
> Ingo Molnar <[EMAIL PROTECTED]> writes:
> > > > it does work for me (i do not actually use it as such, but i\'ve done
some
> > > > stresstesting under heavy load). Let me know if you find any problems.
> > > Hmm? Since when does swapping work on rai
you cant boot off of a raid partition. lilo needs to find the kernel on the
disk using bios routines. raid is not in bios.
you need to create a ~20 meg partition to hold the kernel, etc.
this works very well at redhat install time to create a /boot and /boot2
folder on the first and second disk
have you considered an upgrade to 2.2.x ? i would be interested to see if
problem persists. we are using newest raid code on several uni-proc boxes, no
problems, even with very heavy load. we only using old raid code on SMP boxes.
had considered upgrading to 2.2.x and new raid code on these.
all
have seen what you describe several times, if i have the identical pair of HDs
as master on each of my ide channels. they are set the same in bios, linux
says they are the same at bootup, but fdisk does not. i upgrade bios, this
help one machine, but not most.
i now give the kernel boot time ar
try ftp.XX.kernel.org where XX=your country code, i.e \"us\".
look in the /pub/linux/daemons/raid/alpha dir.
you should patch kernel with the latest avail \"raid0145\" patch from there, and
use a raidtools of similar date from there. use jakob\'s raid-howto that comes
with the latest raid tools.
recently built a pair of systems sim. to your specs.
pII 350
128meg
2 8gig IBM
RH 5.2 and kernel 2.2.1,
primary use as a samba server (logon controller, home dirs, printing
for ~60 win95 clients)
we use the raid0145 1999 01 28 patch.
the array is created with the chunk size set. it is needed.
you are wrong about not being able to mirror / you need the latest alpha raid
patches for this. as well as the latest raid tools from your local
ftp.XX.kernel.org mirror.
see the archives of this list, and esp the documentation that comes in the
latest raid 19990309 patches.
are you familiar wi
i will reply simply because i recently said hardware raid was too expensive.
i said poor private college. every nickel counts here. then you said 17 thousand
dollars. i laughed. apparently you did not read the rest of my post where i said
i had ~2000 users. not 10,000. when it counts, i buy more
you asked for experiences:
pfeiffer university uses linux for nfs, smb, mail and web services. we wanted
security of our data, but could not afford raid controllers. so,
we have been using a redhat 5.0 box with 2.0.35 or 36 kernel and the ancient
raid tools (.40) on three different boxes. combine
you REALLY dont want to build a raid across the same physical harddrive. this is
a bad idea in terms of performance, and what it could do to the drives life
(just think of the heads popping back and forth between two partitions for every
chunk written).
if you can get a second disk in there, the
have seen this problem since the first 2.0.36 patches. i usually make the arrays
with the part. type set to 83, then WITHOUT rebooting, i setup the fs, move the
data, THEN fdisk again, change part. type to fd. then i reboot. this has worked
on a couple boxes, 2.2.1, and 2.0.36. another alternativ
there have been many reports of this. supposedly andrea arcangeli has patches
which fix this. i suggest you look in the archives, cause there was a patch that
fixes this for 2.2.2 posted last week or the week before.
it is a buffering problem with the newer kernels. you could make your arrays
22 matches
Mail list logo