rcpt@eutrc4.urc.tue.nl (Piet Tutelaers) (12/12/89)
We get regular the message `no core' on our uVAXII since we have recently
upgraded our memory capacity from 5 to 13 MB. We use two swap disks,
the default on device /dev/ra0b and an extra on ra1b. The line in /etc/fstab
is:
/dev/ra1b::sw:0:0:ufs::
The partition data of both swap disks (RD53 and RD54) is:
RD53: /dev/rra0a
partition bottom top size overlap
b 15884 49323 33440 c,e
RD54: /dev/rra1a
partition bottom top size overlap
b 15884 49323 33440 c
Vmstat shows the following data:
procs memory page disk faults cpu
r b w avm fre re at pi po fr de sr r1 r2 r0 x3 in sy cs us sy id
0 1 0 1705 4931 0 0 0 0 0 0 0 0 0 1 0 11 16 3 10 5 86
0 0 0 1144 4866 0 0 0 0 0 0 0 0 0 0 0 2 20 1 1 0 99
I have noticed that when the number of free pages is less than 3000 (column
fre), I cannot start up SliTeX for example.
a) Does anyone have suggestions how we can reconfigure the uVAXII in order to
get more programs started?
b) Is it possible to see on a running system which swap devices are active?
c) Is it possible to find out what memory claim a program makes?
Thanks for any help! Piet
------------------------------------------------------------------
uucp: rcpt@urc.tue.nl | Piet Tutelaers Room RC 1.82
bitnet: rcpt@heithe5.BITNET | Eindhoven University of Technology
phone: +31 (0)40 474541 | P.O. Box 513, 5600 MB Eindhoven, NLalan@shodha.dec.com ( Alan's Home for Wayward Notes File.) (12/13/89)
In article <1317@eutrc3.urc.tue.nl>, rcpt@eutrc4.urc.tue.nl (Piet Tutelaers) writes: > > We get regular the message `no core' on our uVAXII since we have recently > upgraded our memory capacity from 5 to 13 MB. We use two swap disks, > the default on device /dev/ra0b and an extra on ra1b. The line in /etc/fstab > is: > /dev/ra1b::sw:0:0:ufs:: > > The partition data of both swap disks (RD53 and RD54) is: > RD53: /dev/rra0a > partition bottom top size overlap > b 15884 49323 33440 c,e > RD54: /dev/rra1a > partition bottom top size overlap > b 15884 49323 33440 c Every looks as expected so far... > > Vmstat shows the following data: > procs memory page disk faults cpu > r b w avm fre re at pi po fr de sr r1 r2 r0 x3 in sy cs us sy id > 0 1 0 1705 4931 0 0 0 0 0 0 0 0 0 1 0 11 16 3 10 5 86 > 0 0 0 1144 4866 0 0 0 0 0 0 0 0 0 0 0 2 20 1 1 0 99 In almost every example I've seen "out of core" is due to a lack of swap space not a lack of memory. After running out of swap space the next limit usually the "datasize" and "stacksize" limits enforced by the C-shell. Rather than vmstat(1) look at "pstat -s". This prints the usage of swap space. The C-shell has a builtin command called limit. You can use it look at what the various limits are. Generally the stack is limited to 512 KBytes. You can use "unlimit" to remove the limit. The limit on "datasize" is controlled by the configuration parameters "dmmin" and "dmmax". They are sufficiently described in the "Guide to System Configuration File Maintenance". > > I have noticed that when the number of free pages is less than 3000 (column > fre), I cannot start up SliTeX for example. > > a) Does anyone have suggestions how we can reconfigure the uVAXII in order to > get more programs started? Determine which of the common two problems is occuring. 1. If it's swap space get more or get rid of an application that is using a lot. 2. If it's the "datasize" or "stacksize" limit, raise them. See what I said before about "dmmin" and "dmmax". > > b) Is it possible to see on a running system which swap devices are active? Using adb you can look at a data structure called _swdevt. The format of this structure is defined in <sys/conf.h>. > > c) Is it possible to find out what memory claim a program makes? You can use size(1) to find out what the startup memory requirements are. The text size limit in V3.x is either 6 MB or 12 MB. In V3.x it might be something that can be configured. > > Thanks for any help! Piet > ------------------------------------------------------------------ > uucp: rcpt@urc.tue.nl | Piet Tutelaers Room RC 1.82 -- Alan Rollow alan@nabeth.enet.dec.com
pavlov@canisius.UUCP (Greg Pavlov) (12/18/89)
In article <1317@eutrc3.urc.tue.nl>, rcpt@eutrc4.urc.tue.nl (Piet Tutelaers) writes: > > We get regular the message `no core' on our uVAXII since we have recently > upgraded our memory capacity from 5 to 13 MB. We use two swap disks,...... > I assume that you modified your config file and recompiled the kernel... > > c) Is it possible to find out what memory claim a program makes? > If you execute "ps -agux" (or many other variants of parameters to ps), you will see a column labeled "SZ". This will tell you how much swap space has been set aside for each process (whether or not the processes will ever "use" all of it). greg pavlov, fstrf, amherst, ny