[hylafax-users] HylaFAX failing my abuse test

[HALLOWELL,KARL (HP-Cupertino,ex1) <karl_hallowell@hp.com>]

My HylaFAX setup uses 4.1beta2 and is running on Redhat 6.2. Under normal
circumstances, it works just fine. I am writing a Java program that uses
gnu-hylafax (http://www.net-foundry.com/java/gnu/hylafax/). I'm just
interested in sending faxes, but I don't recall what configuration choices I
might have made because of that. Can't say much about the Java program
(since my company owns it), but I wrote a nasty stress test to see how well
the code would stay together.

First, I disabled the fax/modem by disconnected the attached phone line
(turning the modem off is possible too, but I didn't do that). Then I start
a Java program that creates ten threads which hammer at my code (and at
HylaFAX) by alternately sending a fax and removing that same fax as fast as
possible. Everything is running on the same machine. All goes well until
hfaxd dies on me. That generates some SocketExceptions (this happens when
the socket connection goes down) and pages of secondary exceptions (Java is
a wondrous thing :-). Unfortunately, no errors appear in any of my system
logs. My hfaxd does log to /var/log/messages, but the messages aren't
enlightening. Around the time that hfaxd dies, I see a number of "CAUGHT
SIGNAL 11" messages from secondary hfaxd processes, but nothing from the
main one. According to my man pages, signal number 11 is "invalid memory
reference" or "SIGSEGV".

My search hasn't uncovered anything on this in the mailing list. Of course,
that doesn't mean it's not there. This bug isn't a serious concern for me,
but the HylaFAX community probably should know about it. Good day!


Karl Hallowell          karl_hallowell@hp.com


____________________ HylaFAX(tm) Users Mailing List _______________________
 To unsub: mail -s unsubscribe hylafax-users-request@hylafax.org < /dev/null