On 07/09/2015 12:22 AM, Scott Bell wrote:
Hi all,
We're experiencing a crash on our web app on a daily basis with
the following message:
Seg fault (internal error) at 0x81a728000
gdb -c racket.core doesn't offer much insight:
Core was generated by `racket'.
Program terminated with signal 6, Aborted.
#0 0x000000080146b6ca in ?? ()
(gdb) bt
#0 0x000000080146b6ca in ?? ()
#1 0x0000000801540179 in ?? ()
#2 0x00000000000188ec in ?? ()
#3 0x0000000000000000 in ?? ()
This Racket 6.1.1 was built on FreeBSD 10.1 from ports with a
small GC patch from rtm.
If I rebuild the port using WITH_DEBUG=yes, which adds -g to
CFLAGS, at least, I get a racket binary that takes significantly
longer to load and compile the app, and still doesn't seem to
have any debug symbols. I assume this is because of the way
the binary is built.
Probably the port doesn't disable the stripping of the symbols. Try
adding "--disable-strip --enable-noopt" to the configure args.
Also, you could give a try to the original port (i.e. without
WITH_DEBUG) with the limits (ulimit/login class) amply increased.
How should I proceed to diagnose the crash? I can modify the
build we use in production, but can't afford more than a minor
(<15%) decrease in performance.
Scott
--
You received this message because you are subscribed to the Google Groups "Racket
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
