Wierd filesystem crash - need help please

I have been experiencing a system crash with QNX 6.0 patch B which is
hard to pin down. At seemingly random intervals (while my application
is running), the filesystem seems to go away. An existing telnet
session (this is an embedded sytem with no keyboard) will give the error
“sh: ls: No such file or directory” if I issue the ls command. Any file
access attempts from my application get the same error. If I try to
start a new telnet session I get the error “telnet: Connection closed by
foreign host”. If I run spin on a telnet session (before the crash) I
don’t see any processes go down when it happens, everything looks normal
(but I don’t really know what to look for). The only way to get the
system back is to cycle the power. When it comes back up everything is
fine. When I look in /var/dumps there isn’t anything unexpected there.

Anybody have any idea what this is all about? It is really becoming an
untenable situation, but I have no idea where to look for the problem.
I have run chkfsys, everything seems ok. Does it have something to do
with the package filesystem? What do I check to find out? Thanks.

I’ve seen this happen when, by mistake, I’ve placed one or more executables
in the wrong section of my build file, along with the shared libraries. By
placing all executables in the [data=c] section this no longer happened.
Basically what was occurring was that an executable would run only once
(e.g. a devb-xxxx filesystem driver).

Just a thought…


“Bruce Davis” <bruce.r.davis@boeing.com> wrote in message
news:3BE2FDAB.DCB63042@boeing.com

I have been experiencing a system crash with QNX 6.0 patch B which is
hard to pin down. At seemingly random intervals (while my application
is running), the filesystem seems to go away. An existing telnet
session (this is an embedded sytem with no keyboard) will give the error
“sh: ls: No such file or directory” if I issue the ls command. Any file
access attempts from my application get the same error. If I try to
start a new telnet session I get the error “telnet: Connection closed by
foreign host”. If I run spin on a telnet session (before the crash) I
don’t see any processes go down when it happens, everything looks normal
(but I don’t really know what to look for). The only way to get the
system back is to cycle the power. When it comes back up everything is
fine. When I look in /var/dumps there isn’t anything unexpected there.

Anybody have any idea what this is all about? It is really becoming an
untenable situation, but I have no idea where to look for the problem.
I have run chkfsys, everything seems ok. Does it have something to do
with the package filesystem? What do I check to find out? Thanks.

Bruce Davis wrote:

I have been experiencing a system crash with QNX 6.0 patch B which is
hard to pin down. At seemingly random intervals (while my application
is running), the filesystem seems to go away. An existing telnet
session (this is an embedded sytem with no keyboard) will give the error
“sh: ls: No such file or directory” if I issue the ls command. Any file
access attempts from my application get the same error. If I try to
start a new telnet session I get the error “telnet: Connection closed by
foreign host”. If I run spin on a telnet session (before the crash) I
don’t see any processes go down when it happens, everything looks normal
(but I don’t really know what to look for). The only way to get the
system back is to cycle the power. When it comes back up everything is
fine. When I look in /var/dumps there isn’t anything unexpected there.

Anybody have any idea what this is all about? It is really becoming an
untenable situation, but I have no idea where to look for the problem.
I have run chkfsys, everything seems ok. Does it have something to do
with the package filesystem?

Sound like a crash of that package filesystem …

Try to copy most of the shell commands to a seperate directory and
include it into the search PATH … so you have a change to operate the
system after a fs-pkg crash (restart of the package filesystem ?)

Armin


What do I check to find out? Thanks.