Broken pipes and vitual memory problems ? HELP

Broken pipes and vitual memory problems ? HELP

Post by Le » Thu, 26 May 1994 14:00:38



WHen I recompile the kernel the first time everything went well.  But when
I recompile after that GCC will stop part way through and report "Out
of virtual memory"  I have 8 megs of system ram a 16meg+ swap space and
lots of free hard drive.

        I also get "broken pipe" messages.  I get these usually when I
use tar with the -zxvf options.  The problem here is when I install
new packeges I think this is the way slackware uncompresses and untars
the new packages.  I usually only happens on big files 500k+.

        Any Ideas ???

        Thanks


 
 
 

Broken pipes and vitual memory problems ? HELP

Post by Bill Hog » Fri, 27 May 1994 14:34:07


: WHen I recompile the kernel the first time everything went well.  But when
: I recompile after that GCC will stop part way through and report "Out
: of virtual memory"  I have 8 megs of system ram a 16meg+ swap space and
: lots of free hard drive.

:       I also get "broken pipe" messages.  I get these usually when I
: use tar with the -zxvf options.  The problem here is when I install
: new packeges I think this is the way slackware uncompresses and untars
: the new packages.  I usually only happens on big files 500k+.

:       Any Ideas ???
        ~~~
 1. free > upload.note
 2. cat /etc/fstab >> upload.note
 3. upload 'upload.note'.
--
  Bill Hogan


 
 
 

1. HELP - Spurious Broken Pipe Problem?

Hello,

We have a program that is running under inetd on a SCO 3.2.4.2 system.
Normally, this program gets started when another (non-SCO Unix) machine
connects to it (by inetd), and the program just stays connected.

We are noticing that occasionally, the program on the other end is getting
a "broken pipe" error.  As I understand it, this occurs when a program
attempts to write to a socket where the reading end has died or closed the
connection.

However, as far as we can tell, the program on the reading/SCO Unix end
has not died, nor closed the socket (under inetd) connection.  [In the SCO
Unix program, there is only one 'close', and it only occurs if the program
exits whereupon it outputs a bunch of log messages...but in the instances
that we're looking at, there were no log messages as such.]

Question:

Has anyone encountered a similar problem?  We are not discounting a
programming error, but barring that, and given that there is only one
place that a close could have occurred in the SCO Unix program, we are
wondering if there are any known 'external' or perhaps non-programmatic
(e.g., configuration errors, etc.) that can cause such "spurious"
occurrences of the 'broken pipe' error?

For example, can things like network traffic, or something like that cause
this error?

Does anyone have any suggestions as to how we might try to track this
problem down?

Thx in advance,
Jim Lum

2. Acerview monitor and 1024x768, win95 works, linux doesn't

3. Help ** XIO Broken pipes problem with X.

4. US-NY-NYC Oracle DBA PeopleSoft Unix

5. problems with tar: broken pipe, Problems with lpd

6. problems with libtermcap.so.2

7. Help! Broken Pipe

8. lvm in 2.5.1

9. help fix broken pipes

10. Help with EPIPE - broken pipe error ...

11. Need help with dump/restore on Linux - Getting broken pipe.

12. Help with Slackware X/3.1.2 please (Broken pipe?)

13. unknown printer...broken pipe...help