rtp1@quads.uchicago.edu (raymond thomas pierrehumbert) (03/28/91)
OK, so I did a ps and found that my problem was inits growing without bound. The two init processes had grown to 32MB each, on a 16MB machine. What do I do about it? Do I kill the processes? How do I keep them from coming back? Does the problem go away at 10.3p? Thanks.
rtp1@quads.uchicago.edu (raymond thomas pierrehumbert) (03/28/91)
Help! My /etc/init processes on my DN10k are growing without bound. They reach 15MB after six hours. I checked my /etc/ttys file, and found no serial line tty's turned on, so I don't think that's the problem. I remember seeing a fix posted here, but ignored it because I though it wouldn't happen to me if it hadn't happened so far. Well it has! So could some kind soul help me out? The machine has become completely unusable.
rtp1@quads.uchicago.edu (raymond thomas pierrehumbert) (03/29/91)
I definitely have the runaway init problem, but it seems to be of a different flavor from that previously reported. My /etc/ttys file has no spuriously "on" lines (Nothing is attached to any of the serial ports, and they are all "off" so far as I can see> My process list and my /etc/ttys file follows. Anybody notice anything? Notice that I seem to have TWO init processes running, which is strange. Notice also that the inits have grown to 23MB (after about 24 hours). % ps -uax USER PID SZ RSS TTY STAT TIME COMMAND root 123040 24 ? S < 1:16 /etc/init root 2 0 0 ? R 1012:28 null root 3 0 0 ? S 0:28 purifier root 4 0 0 ? S 0:00 purifier root 5 0 0 ? S 0:32 unwired_dxm root 6 0 0 ? S 0:00 pinger root 7 0 0 ? S 0:26 netreceive root 8 0 0 ? S 0:05 netpaging root 9 0 0 ? S 0:24 wired_dxm root 10 0 0 ? S 0:01 netrequest root 276 5632 39 ? S 3:00 /etc/tcpd root 284 6144 7 ? S 0:00 /etc/inetd root 287 7168 0 ? S 0:00 /etc/ncs/llbd root 289 8704 22 ? S 0:57 /etc/ncs/glbd root 292 7168 59 ? S < 1:08 /etc/rgyd user 295 6144 0 ? S 0:00 /etc/portmap root 297 6144 0 ? S 0:03 /etc/mountd rtp1 299 8192 0 ? S 0:18 /etc/nfsd sys_pers 304 9728 0 ? S < 0:01 dm root 438023040 732 ? S 0:00 /etc/init root 4317 7680 10 ttyp0 S 0:03 telnetd yang 4318 5632 58 ttyp0 S 0:01 -csh yang 437310240 296 ttyp0 R 1:18 advec41 root 4363 7680 39 ttyp1 S 0:00 telnetd rtp1 4364 5632 38 ttyp1 S 0:00 -csh rtp1 4381 6656 60 ttyp1 R 0:00 ps -uax % more /etc/ttys # # ttys - terminal initialization data # #device getty/program term on/off other flags comment console "/etc/dm_or_spm" apollo on secure # use mkcon to r edirect console output display none apollo off secure # DM pad devices tty01 none dumb off secure tty02 none dumb off secure tty03 none dumb off secure ttyp0 none dumb on secure ttyp1 none dumb on secure ttyp2 none dumb on secure ttyp3 none dumb on secure ttyp4 none dumb on secure
dbfunk@ICAEN.UIOWA.EDU (David B Funk) (03/29/91)
In posting <1991Mar28.165338.14067@midway.uchicago.edu> raymond thomas pierrehumbert writes: > I definitely have the runaway init problem, but it seems to be > of a different flavor from that previously reported. My /etc/ttys > file has no spuriously "on" lines (Nothing is attached to any > of the serial ports, and they are all "off" so far as I can see> > > My process list and my /etc/ttys file follows. Anybody notice > anything? Notice that I seem to have TWO init processes running, > which is strange. Notice also that the inits have grown to > 23MB (after about 24 hours). [ stuff deleted ] > % more /etc/ttys > # > # ttys - terminal initialization data > # > #device getty/program term on/off other flags comment > > console "/etc/dm_or_spm" apollo on secure # use mkcon to r > edirect console output > display none apollo off secure # DM pad devices > tty01 none dumb off secure > tty02 none dumb off secure > tty03 none dumb off secure > ttyp0 none dumb on secure > ttyp1 none dumb on secure > ttyp2 none dumb on secure > ttyp3 none dumb on secure > ttyp4 none dumb on secure Raymond, your "ttys" file is causing your problem. Do NOT turn "on" ports enabled for telnet/rlogin (the "ttyp?" things). These are managed by "/etc/inetd" and not "/etc/init". When you turn them "on" you have both "init" and "inetd" fighting over them, thus the runaway init problem. The only reason for putting "ttyp?" pseudo-ttys in your "ttys" file is to flag them as "secure". Otherwise leave them out and just let inetd take care of them. Here's what you want your ttys file to look like: % more /etc/ttys # # ttys - terminal initialization data # #device getty/program term on/off other flags comment console "/etc/dm_or_spm" apollo on secure # use mkcon to r edirect console output display none apollo off secure # DM pad devices tty01 none dumb off secure tty02 none dumb off secure tty03 none dumb off secure ttyp0 none dumb off secure ttyp1 none dumb off secure ttyp2 none dumb off secure ttyp3 none dumb off secure ttyp4 none dumb off secure Dave Funk