Barton Robinson wrote: > Drop another 20,000 pages, and there > are 40,000 page reads (and 20,000 page writes becuase it was a > constrained storage experiment).
Now that is rather interesting - that instance where the number of reads is exactly twice the number of pages being released maps rather well to the case I mentioned of an 8K PGMBK needing to be read in order to release a 4K data page. So I'm fairly certain the PGMBK reads are part of the issue, I don't see how pages read could be greater than pages released otherwise (assuming this was a controlled experiment with no other factors invovled). But I'm still betting on Linux having the links within the pages themselves, as that would generate the sort of reads you're reporting. And some folks here do see your presentations, though I'll admit I'm usually not one of them - but we are listening. Just keep in mind there's little we can do without a PMR or formal new function request. - Bill Holder, z/VM Development, IBM "Much that once was is lost, for none now live who remember it." - Galadriel "He that breaks a thing to find out what it is has left the path of wisdom" - Gandalf (JRR Tolkein) ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
