mlr@goanna.oz.AU (Mark Ross) (05/19/89)
We have two Encore machines in our Department, a 320 running V.2.2.f and a 310 running Umax 4.2 R3.3.0. We are experiencing a problem with Umax 4.2 R3.3.0 that I hope someone may know something about or even better know if a fix is available. The problem is as follows: After the machine has been running for a few hours it begins to kill new processes. It is at about the same time that the paging activity on the system gets very high. The system has about 30 users on it and the load is mostly less than 2. Most users just access email so the work performed by the system is quite trivial. At the time when the paging activity is high the system kills new processes and reports 1) No more processes 2) Out of memory The system is unusable when it is in this state. The only solution is a reboot. The configuration of the 310 system is 1APC,1EMC,16Mbytes and 2 Wren IV drives. Some of the software configurations are Maxusers: 60 Max_procs: 50 Nswbuf: 32 Wssrate: 2000000 Maxwire: 0 There is 40Mbyte of swap space on the system Is anyone else having this problem? On a separate point does anyone know whether a Umax4.3 is going to become available or is the merged UmaxV the direction be developed? Mark Ross, Senior Lecturer ACSnet: mlr@goanna.cs.rmit.oz UUCP: ...!uunet!munnari!goanna.cs.rmit.oz!mlr CSNET: mlr@goanna.cs.rmit.oz ARPA: mlr%goanna.cs.rmit.oz@uunet.uu.net BITNET: mlr%goanna.cs.rmit.oz@CSNET-RELAY PHONE: + 61 3 660 2347 Department of Computer Science, Royal Melbourne Institute of Technology, P.O. Box 2476V, 124 Latrobe St, Melbourne, 3000, Australia ------
reeder@ut-emx.UUCP (William P. Reeder) (05/23/89)
In article <8905181054.AA18446@uunet.UU.NET>, mlr@goanna.oz.AU (Mark Ross) writes: > After the machine has been running for a few hours it begins to kill > new processes. It is at about the same time that the paging activity > on the system gets very high. The system has about 30 users on it > and the load is mostly less than 2. Most users just access email > so the work performed by the system is quite trivial. > > At the time when the paging activity is high the system kills new processes > and reports > 1) No more processes > 2) Out of memory > > The system is unusable when it is in this state. The only solution > is a reboot. The configuration of the 310 system is 1APC,1EMC,16Mbytes and > 2 Wren IV drives. Try adding more memory. We had exactly this problem with a Multimax 320. Encore claimed that we were underconfigured on memory, they now recommend 16MB per APC card (although that is what you have :-). Adding more memory to our system avoided the problem. If you will look, you will probably find that the system is actually taking up close to half of the available physical memory, and is simply refusing to take up any more. (It dynamically allocates its tables.) There is probably virtual memory available, the limit being exceeded is the memory limit for the OS. The system does start paging madly, since there are lot's of processes trying to use not much memory, and UMAX 4.2's memory paging/swapping algorithm is not as good as it could be. Do you see any suspended jobs taking up physical memory? Or have any runable jobs completely swapped out? We did! Of course, our head kernel haquer was also scatching his head wondering why the OS was taking up 15+ megabytes of memory. So there seems to be two problems: why is the OS so bloated? and the VM system is not smart enough. Anyway, adding more memory should help. Maybe Encore will fix this problem in UMAX 4.3. -- Wills reeder@emx.utexas.edu, uunet!cs.utexas.edu!ut-emx!reeder -- DISCLAIMER: Opinions expressed above are those of the author and do not neccessarily reflect the opinions of the manufacturer of the wires which brought it to you.