www.a00.de > tcpgroup > 1992 > msg00221
 

TCP-group 1992


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

wg7j update



> i made a few minor bugfixes to my code (the netrom l4 stuff ala dave perry,
> and other things) The ARP autoroute is now forced off when rspf is used,
> per a good suggestion by Mike Billow .

A few observations:

1. WOW.... nice touch! Smoothed out ALL of my headaches
here with the gateway. Particularly SECURITY!!!

2. Thanks for fixing the ^b problem! (I hate the "smiling
bastard" character.... 8-} ) Adding cls was also nice.

3. Memory Leaks!...... still running into severe sudden
memory "hemorrhage" on my 386sx 16 (Phoenix Bios)... on a
quiet ethernet, a "private" SLIP link, and an AX25 port
that is used by the encap gateway users. OS is mess-doo 5
and sometimes Windows 3.0... (symptoms occur independent of
Windows!)

        -some appears to occur related to the inability of
        the memory code to "recover" surrendered malloc
        after a number of sessions are closed.

        -other observations after a weekend of use... if
        (for whatever reason! ;-} ) a remote session
        through the gateway closes (read: is probably
        dropped by remote user) without a final ack/RST a
        number of close pending and close wait sessions
        stack up. This is almost universally fatal! The
        throughput and response time of the system
        approaches 6802 or worse, then locks up. Net log
        shows no abnormalities through the last log update,
        however, a number of .tmp files are open and left
        that way!

        -recovery from either state by kicking or resetting
        the individual sessions does not improve the memory
        situation for up to 10 hours afterward. As has been
        noted by many users... the direct symptoms are that
        heap continues to grow, and freeheap varies from 8%
        to 20%... while morecores continue to drop to the
        point that a shell invocation is no longer
        possible. The time it takes to reach "critical" is
        on the order of 16 to 48 hours. (BTW.. all code,
        Phil's December release, Gerards 1.8b and c, and
        yours all seems to do this in my environment.

Since I dont have time to "go into the larval stage" with
the code and have barely had time (approaching Ph.D. thesis
defense in Microbial Genetics....|8-[} ) to play with my
compiler, I use the code compiled and distributed by the
author of each particular release. As a "beta" tester, I think
this is a better approach (eliminates compiler and "luser"
errors!!!) anyway because the author has an idea how to
look for a problem in his own development context.

One other comment.. Could you and Gerard "slow down and
catch up with each other"? You are leapfrogging each other
with your stuff because you both are so quick with this
stuff. You indicated you used GRI 1.8a as your basis code
and Gerard has both debbugged it AND added some of your
features to his release 1.8c! Nice place to begin the next
round would be to diff yours and his and then "reunify" at
1.8d or what ever.....

Both of you, keep up the great work, you  have done a nice
job with Phil's latest code!!

73

| jack spitznagel, d.d.s.
| Internet:   jks@giskard.uthscsa.edu
|             spitznagel@thorin.uthscsa.edu
| Bitnet:     spitznagel@uthscsa
| AmprNet:    kd4iz@sat.ampr.org
| CIS:        76044,476





Document URL : http://www.a00.de/tcpgroup/1992/msg00221.php
Ralf D. Kloth, Ludwigsburg, DE (QRQ.software). < hostmaster at a00.de > [don't send spam]
Created 2004-11-12. Last modified 2004-11-12. Your visit 2020-10-28 17:28.55. Page created in 0.0533 sec.
 
[Go to the top of this page]   [... to the index page]