root (11/19/82)
Our "uucico" program have a "bug" which is driving me up the wall, the
problem is this:-
We are currently connected to 4 site on the net, one of the link
is reported to be very noisy, unfortunately this is the link where a huge
volume of data is transfered. This combination of noisy link and mass
volume of data transfer causes the "uucico" process to hang on some
error condition intermittently. (this never happened to the other
three link). Yes, it simply hang and does not time out, and it does
not exit.
The problem with debugging such a beast is that of recreating
the error condition caused by the noisy line. Almost in variably it
works when we are testing it. But when we put it back to production
run, it works for a few days, and the problem started to come back
every now and than.
The failure pattern I have collected so far is as follow:
it seems that it always failed in a the "SLAVE" mode waiting for
for a file to be transfered from the other site. I do not have
enough debugging output yet to locate where the process is hanging,
(stderr is closed when it is started from "cron", To collect
debugging output , I wrote a 4 line program which attach stderr to
"/tmp/uuLOG" , then
"execl("/usr/lib/uucp/uucico", "uucico", "-r1", "-x7",0);
now is a matter of waiting for it to fail again.
)
Since "/tmp/uuLOG" has not recorded any failure yet, I can only
make a guess base on the information in "../spool/uucp/LOGFILE".
It seem that "uucico" usually hang somewhere in the routine
"pkread(S)" in pk0.c . In fact, I am suspicious about
the line such as
"while (pkaccept(pk) == 0) ;"
Or is the "time out" done somewhere else ?? (i.e gio.c or pk1.c ??)
Does anyone had similar problem or/and can anyone provide some
information on this problem ??
please mail reply to ..!allegra!sbcs!andrew
or ..!peri!sbcs!andrew
Andrew W. Chang
Dept. of Computer Science
State University of New York