carsontr@utcsri.UUCP (Carson T. Schutze) (01/12/88)
Sorry if one of these groups isn't the right one for this query, but I'm
not too clear on what the difference is.
Christopher Schmidt recently sent us a version of TCP/IP to use under
Koto2.0 to communicate with a Sun (3/280). We have written a program, to be
used as a print daemon, which talks to the Sun every minute, checking for
new files in a spool directory. The program seems to work correctly, but
when we leave this program running, within 2 days the 8MB and Vmem memory
indicators read completely full, and the machine grinds to a halt.
We had a similar program that used PUP and did not suffer the same
problem. Does anyone know if TCP leaves lots of garbage around, or why this
might be happening?
---------------------------------------------------------------------
Carson T. Schutze Dynamic Graphics Project
Computer Systems Research Institute
(416) 978-6619 University of Toronto
Toronto, Ontario, Canada M5S 1A4
BITNET: carsontr@explorer.dgp.utoronto
EAN: carsontr@explorer.dgp.toronto.cdn
UUCP: ...!mcvax!explorer.dgp.toronto.edu!carsontrbkn@ida.liu.se (Bernt Nilsson) (01/16/88)
We have very nearly the same problem: We wrote a TCP print service in Interlisp, that takes requests from TCP and sends them to our Xerox printers. Because of the nature of the requests (first comes the files, and last the request it self) we have to have a local spool directory. The machine we use for these kind of lisp service programs is an old 1108 with only 10MB disk. Therefore we use CORE-files for spooling. Everything works OK except that the vmem slowly gets full, so we have to install a fresh sysout after a couple of days (can't use VMEM.PURE.STATE either). Our old guess was that it is the CORE-files that don't get garbage collected correctly. But maybe it is TCP that is the problem. -- Bernt -------