gamiddle@watmath.waterloo.edu (Guy Middleton) (04/10/91)
We're having some trouble with our RC6280. If anybody has a stable configuration (one that stays up for more than one day), I would very much like to hear from you, especially if you have any advice. Thanks. -Guy Middleton, University of Waterloo gamiddleton@watmath.waterloo.edu (+1 519 885 1211 x3472) gamiddleton@watmath.uwaterloo.ca
at@cc.tut.fi (Toivo Veli) (04/10/91)
In article <1991Apr9.202253.13384@watmath.waterloo.edu> gamiddle@watmath.waterloo.edu (Guy Middleton) writes: > We're having some trouble with our RC6280. If anybody has a stable > configuration (one that stays up for more than one day), I would very much > like to hear from you, especially if you have any advice. Thanks. We have one, which seems to have been stable for about 2 weeks now. Our machine crashed regularly after staying up for about 2 - 4 days, but now it has broken the uptime record by doubling the last value (a week, made in a holiday season): # uptime 1:11pm up 14 days, 3:31, 55 users, load average: 3.73, 3.22, 2.67 # How did we reach this? Simple, just configure the kernel parameters NBLK4096 & NBLK2048 from 2 to 16.... How did we know which parameters to change? Simple, our hardware maintenance was in contact with Mips (UK) more than our local distributor... We have now found a few more interesting features, which can be seen only when the machine runs long enough. Talkd and in.phoned both seem to be hanging around in process table, so I have now killed about 50 of each. Our /etc/passwd now contains 1760 lines and 600 - 700 of our users visit the machine daily. Average simultaneous users has been as high as 96, but now seems to be around 60 - 70 at maximum. We are still running RiscOS 4.51, but as soon as 4.52 comes here, we try to get it. We heard that these diskquota-panics have been somewhat cleaned in the new version. We have had plenty of them... The machine was installed last year. If you have any questions I will try to answer. - at -- Anssi Toivo - TUT Computer Centre - at@cc.tut.fi
rogerk@mips.com (Roger B.A. Klorese) (04/11/91)
In article <AT.91Apr10142245@lehtori.cc.tut.fi> at@cc.tut.fi (Toivo Veli) writes: >We are still running RiscOS 4.51, but as soon as 4.52 comes here, >we try to get it. We heard that these diskquota-panics have been >somewhat cleaned in the new version. We have had plenty of them... Until you receive 4.52, a workaround for the diskquota panics is to increase the size of the disk quota cache: /etc/kopt set _riscos_max_disk_quota_cache 512 (It starts out as 128 for RC6280s.) You can also do this by binary reconfiguration, changing the value of RISCOS_MAX_DQUOT_CACHE in /usr/reconfig/master.d/kernel.<cpu-board>_std. -- ROGER B.A. KLORESE MIPS Computer Systems, Inc. MS 6-05 930 DeGuigne Dr. Sunnyvale, CA 94088 +1 408 524-7421 "10 years of Reagan/Bush have brought us to a new place: postconstitutional America." - Jon Carroll rogerk@mips.COM | {ames,decwrl,pyramid}!mips!rogerk
gamiddle@watmath.waterloo.edu (Guy Middleton) (04/11/91)
I would like to apologise for the negative tone in my last posting. I have just been informed that our hardware people and the MIPS local office are still working on the problem, and I really should not have bothered the net with this at the moment. Sorry... -Guy Middleton, University of Waterloo gamiddleton@watmath.waterloo.edu (+1 519 885 1211 x3472) gamiddleton@watmath.uwaterloo.ca
at@cc.tut.fi (Toivo Veli) (04/11/91)
In article <2138@spim.mips.COM> rogerk@mips.com (Roger B.A. Klorese) writes: > Until you receive 4.52, a workaround for the diskquota panics is to > increase the size of the disk quota cache: > > /etc/kopt set _riscos_max_disk_quota_cache 512 > > (It starts out as 128 for RC6280s.) Straight from our local start-up file: ... echo "Fixing bug in quota-allocation" > /dev/console /etc/kopt set _riscos_max_disk_quota_cache 512 ... Didn't make any difference. We have not tried to configure this as default into the kernel, but it shouldn't cause any difference, should it? - at -- Anssi Toivo - TUT Computer Centre - at@cc.tut.fi Tampere University of Technology
zdenko@katzo.rice.edu (zdenko tomasic) (04/11/91)
In article <AT.91Apr11144400@lehtori.cc.tut.fi> at@cc.tut.fi (Toivo Veli) writes: > >In article <2138@spim.mips.COM> rogerk@mips.com (Roger B.A. Klorese) writes: >> Until you receive 4.52, a workaround for the diskquota panics is to >> increase the size of the disk quota cache: >> >> /etc/kopt set _riscos_max_disk_quota_cache 512 >> >> (It starts out as 128 for RC6280s.) > >Straight from our local start-up file: >... > echo "Fixing bug in quota-allocation" > /dev/console > /etc/kopt set _riscos_max_disk_quota_cache 512 >... > >Didn't make any difference. We have not tried to configure this ^^^^^^^^^^^^^^^^^^^^^^^^^^^ It is my impression that some kopts settings do not take effect unless set at the monitor prompt (i.e. before system boot) >as default into the kernel, but it shouldn't cause any difference, >should it? > > - at >-- >Anssi Toivo - TUT Computer Centre - at@cc.tut.fi >Tampere University of Technology -- ___________________________________________________________________ Zdenko Tomasic, Rice U., Chem. Dept., P.O. Box 1892, Houston, Tx 77251 INTERNET: zdenko@katzo.rice.edu ___________________________________________________________________
rogerk@mips.com (Roger B.A. Klorese) (04/12/91)
In article <1991Apr11.153215.15451@rice.edu> zdenko@katzo.rice.edu (zdenko tomasic) writes: >It is my impression that some kopts settings do not take effect unless >set at the monitor prompt (i.e. before system boot) kopt is *never* run at the monitor prompt. It is a Unix command, and should be run as you did. -- ROGER B.A. KLORESE MIPS Computer Systems, Inc. MS 6-05 930 DeGuigne Dr. Sunnyvale, CA 94088 +1 408 524-7421 "10 years of Reagan/Bush have brought us to a new place: postconstitutional America." - Jon Carroll rogerk@mips.COM | {ames,decwrl,pyramid}!mips!rogerk
rogerk@mips.com (Roger B.A. Klorese) (04/12/91)
In article <AT.91Apr11144400@lehtori.cc.tut.fi> at@cc.tut.fi (Toivo Veli) writes: >Straight from our local start-up file: >... > echo "Fixing bug in quota-allocation" > /dev/console > /etc/kopt set _riscos_max_disk_quota_cache 512 >... >Didn't make any difference. We have not tried to configure this >as default into the kernel, but it shouldn't cause any difference, >should it? > No, it shouldn't. This may not, in fact, make any difference. It doesn't cause crashes not to happen; it merely delays the time the cache will wrap around and get corrupted. -- ROGER B.A. KLORESE MIPS Computer Systems, Inc. MS 6-05 930 DeGuigne Dr. Sunnyvale, CA 94088 +1 408 524-7421 "10 years of Reagan/Bush have brought us to a new place: postconstitutional America." - Jon Carroll rogerk@mips.COM | {ames,decwrl,pyramid}!mips!rogerk