This has not been fixed in 2.1.42 at all. I need to confirm that on some
machines (I think the issue is slow disks) mke2fs never finishes, no matter
how much memory you give, provided that it is less than needed.
I have a SS5 with normally 48M of RAM, now 80M, where am I trying to make a
9G filesystem on 3 disks put together into raid0. No matter of what kernel
I'm using, it locks up after it eats all available memory by buffers.
I've tried 2.0.30 and 2.1.42. I have also tried to open the device with
O_SYNC, it died as well, but I could see on the other vt how buffers are
growing and are never discarded, eventhough they must have been saved to
disk due to O_SYNC. In the final stage, I had about 70M in buffers. Then I
couldn't run any single program - all memory was eaten and nobody was
deciding to discard them... The machine had 100M of free swap all the time
(but that's not an issue - it shouldn't be swapping anything).
On the other hand, I need to report success of making a 47G raid0 on a SS20
with 288M of RAM (but for such volume the buffers should be larger than
physical RAM as well).
Is anyone going to look into this or should I have a look at it?
Cheers,
Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
Ultralinux - first 64bit OS to take full power from the UltraSparc
Linux version 2.0.30 on a sparc machine (291.64 BogoMips).
___________________________________________________________________