Обсуждение: (OT) Linux limits
Sorry for the off-topic question, but... I've got a (laptop) system running Mandrake 6.1 which is configured out of the box to disallow core dumps from users. root is allowed to increase the size limit (from tcsh, use "limit coredumpsize unlimited") but users are not allowed to do this for themselves. Where does one specify this parameter on a system-wide basis? My older RedHat boxes all have a non-zero limit for this parameter, and allow setting the limit to infinity by users. Don't know if Mandrake is configured differently from RH6.1, but until I get this adjusted it doesn't make a reasonable development machine... - Thomas -- Thomas Lockhart lockhart@alumni.caltech.edu South Pasadena, California
Hello Thomas, > Where does one specify this parameter on a system-wide basis? My older > RedHat boxes all have a non-zero limit for this parameter, and allow > setting the limit to infinity by users. Don't know if Mandrake is > configured differently from RH6.1, but until I get this adjusted it > doesn't make a reasonable development machine... I have Mandrake 6.0 though I think I know your problem. In /etc/profile, there's an entry there where it limits core dumps to 100000 I think for root. You might want to remove that or make it unlimited. Regards, Neil D. Quiogue STO - dotPH, Inc. "Nothing great was ever achieved without enthusiasm." - Ralph WaldoEmerson
> > I've got a (laptop) system running Mandrake 6.1 which is configured > > out of the box to disallow core dumps from users. root is allowed to > > increase the size limit (from tcsh, use "limit coredumpsize > > unlimited") but users are not allowed to do this for themselves. > are you looking for /etc/security/limits.conf ? Thanks for the tip, and it looks like the right thing, but adding entries for core and rebooting does not help. I then tried upping a brute-force limit of zero imposed in the daemon startup function in /etc/rc.d/init.d/functions thinking that inetd or loginout or somesuch process might need to be higher (since all children inherit these limits apparently), but that does not seem to help. It is set to zero in /etc/profile, and commented out in /etc/csh.cshrc, but afaik anything set at that point should be able to be set higher later. There is a cryptic comment in /etc/profile saying that "for bash2 it can't be set higher for user processes", but I don't know what that's about. Can someone running a Mandrake6.1 or RH6.1 system take a look at their system limits (for csh use "limit", for bash use "ulimit -a"). Are they greater than zero for the coredumpsize?? - Thomas -- Thomas Lockhart lockhart@alumni.caltech.edu South Pasadena, California
Thomas Lockhart wrote: > setting the limit to infinity by users. Don't know if Mandrake is > configured differently from RH6.1, but until I get this adjusted it > doesn't make a reasonable development machine... My experience has been that starting with version 6.0 Mandrake is diverging from RedHat. Mandrake 5.3 can properly be called 'RedHat 5.2+KDE+enhancements' -- Mandrake 6.0 and 6.1, being released before their RedHat counterparts, are not nearly as close. I tried using Mandrake 6.0 to build RPMs, and quickly replaced it with RedHat 6.0 -- Mandrake 6.0 used pgcc instead of egcs, for one. Caused me all manner of grief. Mandrake 6.1 may be better in this regard, but I am sticking with RedHat for the time being, as it is the current baseline target of the RPM distribution. From what I understand, the RedHat binary RPM's still work with Mandrake. Mandrake is now a full-fledged distribution, not just another RedHat knock-off. I'm going to have to get my home machine into a multidevelopment mode, with RedHat, Caldera, SuSE, and Mandrake multibooting, as each of these RPM-based distributions is different, although Mandrake and RedHat are more alike than SuSE and Caldera. Or, you can help me with Mandrake issues in both the source and binary RPM's, just as I am getting assistance from others with the Alpha patches, building/installing the RPM's under SuSE and Caldera, and other architecture (ARM and MIPS come to mind) issues. Portability amongst Linux distributions is becoming nearly as big of issue as portability amongst different Unices. -- Lamar Owen WGCR Internet Radio 1 Peter 4:11
Thomas Lockhart wrote: > Can someone running a Mandrake6.1 or RH6.1 system take a look at their > system limits (for csh use "limit", for bash use "ulimit -a"). Are > they greater than zero for the coredumpsize?? Mandrake 5.3: [lowen@www lowen]$ ulimit -a core file size (blocks) 1000000 data seg size (kbytes) unlimited file size (blocks) unlimited max memory size (kbytes) unlimited stack size (kbytes) 8192 cpu time (seconds) unlimited max user processes 256 pipe size (512 bytes) 8 open files 256 virtual memory (kbytes) 2105343 [lowen@www lowen]$ RedHat 6.0: [lowen@backup lowen]$ ulimit -a core file size (blocks) 1000000 data seg size (kbytes) unlimited file size (blocks) unlimited max memory size (kbytes) unlimited stack size (kbytes) 8192 cpu time (seconds) unlimited max user processes 256 pipe size (512 bytes) 8 open files 1024 virtual memory (kbytes) 2105343 [lowen@backup lowen]$ RedHat 6.1: [lowen@utility lowen]$ ulimit -a core file size (blocks) 1000000 data seg size (kbytes) unlimited file size (blocks) unlimited max memory size (kbytes) unlimited stack size (kbytes) 8192 cpu time (seconds) unlimited max user processes 2048 pipe size (512 bytes) 8 open files 1024 virtual memory (kbytes) 2105343 [lowen@utility lowen]$ Don't have a Mandrake 6.1 system up. -- Lamar Owen WGCR Internet Radio 1 Peter 4:11
On Fri, 7 Jan 2000, Thomas Lockhart wrote: > > > I've got a (laptop) system running Mandrake 6.1 which is configured > > > out of the box to disallow core dumps from users. root is allowed to > > > increase the size limit (from tcsh, use "limit coredumpsize > > > unlimited") but users are not allowed to do this for themselves. > > are you looking for /etc/security/limits.conf ? > > Thanks for the tip, and it looks like the right thing, but adding > entries for core and rebooting does not help. I then tried upping a > brute-force limit of zero imposed in the daemon startup function in > /etc/rc.d/init.d/functions thinking that inetd or loginout or somesuch > process might need to be higher (since all children inherit these > limits apparently), but that does not seem to help. Under FreeBSD, we have a similar file: login.conf ... after modifying it, though, you have to run a command to "compile" it ... do you have something similar?