Re: Oracle8 on Linux ok?

George Dau (gedau_at_isa.mim.com.au) wrote:

: 1. Put Oracle on 4 SCSI disks on the one SCSI bus. Add a CD-ROM and a Tape
: drive. Run a hot backup of the database to the tape while accessing the CD-ROM.
: You run out of SCSI buffers real fast and the whole system freezes (does not

This may be a bug in the SCSI driver. Some of the AIC-7xxx drivers are known to be bad and there are newer drivers available. Maybe this is your problem? This is no worse than dodgy drivers in NT (and there are lots of those!).

Anyway, this is getting too Linux technical for an oracle group

: 2. Mount a non-existant IDE partiton. Wipes the whole thing out straight away.

No. You must have something there. If the partition doesn't exist then trying to access it will give errors, but no crash: On my machine /dev/hda3 doesn't exist.

  mutley.tty3% uname -a
  Linux mutley.XXX.XX.XX 2.0.35 #4 Mon Nov 23 20:04:29 GMT 1998 i686 unknown

  mutley.tty3% mount /dev/hda3 /mnt
  /dev/hda3: Success
  FAT bread failed
  FAT bread failed
  mount: you must specify the filesystem type

  mutley.tty3% mount -t ext2 /dev/hda3 /mnt   EXT2-fs: unable to read superblock
  mount: wrong fs type, bad option, bad superblock on /dev/hda3,

         or too many mounted file systems
         (aren't you trying to mount an extended partition,
         instead of some logical partition inside?)

If you have what looks like a partial filesystem there then the kernel will attempt to mount it and if it's corrupted then you could corrupt the filesystem cache. However, this is plain and simple pilot error in this case. If you tell the system to load corruption, then it will load it! Unix does not stop you from shooting yourself in the foot if that is your desire, and Linux is no different here.

: 3. Put about 90 concurrent users on a Pentium 90 - AIEEEE!! Scheduling in
: interrupt and the system is as dead as an BSOD NT box.

Gleep! That's one seriously overloaded machine! You've probably stress-tested that machine beyond the limits and found a bug. I know the scheduler has been rewritten for 2.2 kernels, so maybe this has been fixed. Heh :-) I've never even try 90 concurrent users on a P90 on any other OS!!!

: That said; I am amazed at how people have reported 1.5 years uptime for the NT

I don't believe that because MS have said that an internal timer rolls over before then (70 odd days I think?) and causes instability. There's a patch for this :-)

--

rgds
Stephen Received on Wed Apr 14 1999 - 01:54:05 CDT