Subject: Re: Alright, damnit, this is getting annoying.
To: Bertram Barth <email@example.com>
From: Johnny Billquist <bqt@Update.UU.SE>
Date: 06/12/1997 19:50:44
> > My current problem is that the system don't swap. I'm searching for the
> > bug right now...
> In the meantime i've upgraded to 1.2F and re-checked both swapping on
> SCSI and swapping on nfs. For my machine it turned out that the machine
> starts swapping reliable, but then it breaks.
What do you mean by "starts swapping"? Do you know for sure that pages are
written, or just that pages have been marked for paging out.
> - on the scsi-variant this might be due to the SCSI-driver, i didn't
> have enough time to go deeper into that yet.
> - on the nfs variant some pages are swapped out, but obviously not enough
> or not fast enough to keep the system running, this might be related to
> the fact that i'm using my uVAX-II as nfs-server for swap. I can see
> some pages geing swapped out, the free-memory decreases to zero, and it's
> the same as before. I don't know if this is really related to the
> slowness of this nfs-server, but i have no other explanation/clues.
It sure sounds like the same problem I'm having.
> Do you have more info about the swap problem? Have you found the reasons?
Yes and no. I've tracked the problem down to a routine in pmap.c, which
hangs in an eternal loop, and I've sent mail to Ragge about it.
I might try to sort it out myself one of these nights...
> PS: What kind of disks are you using which prevent vmstat from working?
I've already found and fixed that problem, and the fix are part of 1.2F.
It turned out that MSCP didn't setup the disk name in the structures, so
when vmstat tried to find the name of the disk, it got a null pointer.
I'm using one RA72, but I would like to add another one, and need some
hardware information before I can do.
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: firstname.lastname@example.org || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol