To my question:
> Shouldn't the paging algorithm spread the I/O across all of the swap
> devices available to balance the load?

Willem Konynenberg replied:

>When you didn't explicitly specify a priority for each swap volume,
>the system automatically assigned priorities in descending order.
>If you want load balancing, specify a priority explicitly.

Thanks. Willem. That fixed the problem. By adding pri=1 to each of the swap
spaces in fstab, I'm getting a more even balance over 6 swap spaces.

Also, when I had unequal swap spaces, Linux S/390 would go into a
supervisor loop over all CP's (probably a spin locks) when the first swap
space filled up. WIth the even distribution, the supervisor loop has been
corrected and we can get on with our work (albeit with a log of paging).

Since swap space is really page space in Linux 2.2 and above, why don't we
call it page rather than swap?

Regards, Jim
Linux S/390-zSeries Support, SEEL, IBM Silicon Valley Labs
t/l 543-4021, 408-463-4021, [EMAIL PROTECTED]
*** Grace Happens ***

Reply via email to