jhl@frith.msu.edu (John Lawitzke) (07/06/90)
One of our customers is having the following problem. I can't rationally explain it. They've contacted SCO support and didn't get an answer. Can anyone out there explain it? Randomly, about once or twice a week on different machines they will get the message from /etc/init: /dev/tty01: getty keeps dying - there may be a problem ANd then tty01 is locked up. They've also seen it happen on tty2c. Now the strange thing, someone is looged in and working on the line when it occurs. There is no getty running on the line and nothing should be starting one on it. They are developing a custom application written in DBL on the machines it is occuring on. My only explanation is that their application is doin something wrong. I've never seen this problem happen internally at my company or at any of our other customers. However, they aren't convinced and keep blaming the hardware or operating system. Any suggestions? -- j |%|John Lawitzke, Dale Computer Corp., R&D |%|UUCP: uunet!mailrus!sharkey!dale1!jhl |%| or: uunet!frith!dale1!jhl Inquiring minds just wondering. |%|Internet: jhl@frith.egr.msu.edu
max@mthvax.cs.miami.edu (Max Southall) (07/07/90)
>Randomly, about once or twice a week on different machines they will >get the message from /etc/init: > >/dev/tty01: getty keeps dying - there may be a problem > Yup, see it all the time, randomly. Happens once or twice a day. And it's on nonexistent ports most of the time, too! When it's a real port, it disables it ... very annoying. This is just with a couple of external ports for dial-in - no custom software! The version I have is Xenix 286 2.2.3 ...
gt0178a@prism.gatech.EDU (BURNS,JIM) (07/07/90)
in article <1990Jul6.171544.10473@mthvax.cs.miami.edu>, max@mthvax.cs.miami.edu (Max Southall) says: > When it's a real port, it disables it ... very annoying. > This is just with a couple of external ports for dial-in - no custom > software! This sounds like an intermittent carrier detect. (Getty would try to re- establish itself in that case, and probably cause a race condition on the port.) Isn't there a timeout value you can set for CD drops? -- BURNS,JIM Georgia Institute of Technology, Box 30178, Atlanta Georgia, 30332 uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt0178a Internet: gt0178a@prism.gatech.edu
gary@cdthq (gary) (07/08/90)
max@mthvax.cs.miami.edu (Max Southall) writes: > >/dev/tty01: getty keeps dying - there may be a problem > Yup, see it all the time, randomly. Check and make sure all the handshaking lines on your serial port are terminated (either looped back to it's opposite number on the back of the computer, or driven by your terminal). Noise on open lines has caused me problems in the past. Gary Heston, at home.... the problems were at work, though...
milan@gpu.utcs.utoronto.ca (Milan Strnad) (07/09/90)
I've seen this happen many times, but only when the particular terminal line was a significant distance away from the actual xenix box. One kludge is to (once in a while) disable the tty and then enable it again. Perhaps SCO has a more elegant solution.
jhl@frith.uucp (John Lawitzke) (07/09/90)
From article <9Z1yL1w161w@cdthq>, by gary@cdthq (gary): $ max@mthvax.cs.miami.edu (Max Southall) writes: $> >/dev/tty01: getty keeps dying - there may be a problem $> Yup, see it all the time, randomly. $ $ Check and make sure all the handshaking lines on your serial $ port are terminated (either looped back to it's opposite number $ on the back of the computer, or driven by your terminal). Noise $ on open lines has caused me problems in the past. No one seems to be reading my original posting correctly. It is happening on /dev/tty01, one of the CONSOLE multiscreens. -- j |%|John Lawitzke, Dale Computer Corp., R&D |%|UUCP: uunet!mailrus!sharkey!dale1!jhl |%| or: uunet!frith!dale1!jhl Inquiring minds just wondering. |%|Internet: jhl@frith.egr.msu.edu
shields@yunccn.UUCP (Paul Shields) (07/10/90)
max@mthvax.cs.miami.edu (Max Southall) writes: >>/dev/tty01: getty keeps dying - there may be a problem >When it's a real port, it disables it ... very annoying. >The version I have is Xenix 286 2.2.3 ... I find this happens if a modem's carrier detect is "on" when I boot Xenix. The driver normally tells you when booting which serial cards are installed and how many ports they have. But if the modem CD light is on, it ignores the card. Getty will die on these "non-existent" ports. I think it's a bug the serial driver. Perhaps there is a fix for this? -- Paul Shields, shields@nccn.yorku.ca (..!uunet!yunccn!shields) Playing devil's advocate, as usual.
max@mthvax.cs.miami.edu (Max Southall) (07/10/90)
In article <1990Jul9.164654.1795@msuinfo.cl.msu.edu> jhl@frith.uucp (John Lawitzke) writes: >From article <9Z1yL1w161w@cdthq>, by gary@cdthq (gary): > >No one seems to be reading my original posting correctly. It is >happening on /dev/tty01, one of the CONSOLE multiscreens. > I understood ... it happens on tty08 to me! as well as on uninstalled serial ports! Maybe the other posters aren't SCO Xenix users and therefore didn't pick up on that ...
rogerk@sco.COM (Roger Knopf 5502) (07/17/90)
In article <1990Jul6.130423.16461@msuinfo.cl.msu.edu> jhl@frith.msu.edu (John Lawitzke) writes: >Randomly, about once or twice a week on different machines they will >get the message from /etc/init: > >/dev/tty01: getty keeps dying - there may be a problem This message comes when getty is forked rapidly many times in a short period of time. Since the usual reason for this behavior is that bogus input is coming in the port, init stops forking getty for a while in the hope that the guy sitting on the keyboard will get up.... Since this isn't a serial port, it isn't junk coming in the modem or anything like that. >ANd then tty01 is locked up. They've also seen it happen on tty2c. Now >the strange thing, someone is looged in and working on the line when it >occurs. There is no getty running on the line and nothing should be >starting one on it. They are developing a custom application written >in DBL on the machines it is occuring on. The thing with tty2c blows my mind, this is not normal at all. Did you somehow get a getty running on the tty while the application was running? (Use "ps -lt tty2c" and see if there is a getty). Something seems not right here - it may be hardware, I seriously doubt it is the application but in any case you will have to be quite rigorous in documenting everything that is happening on the system that might affect console logins. Also, just as a longshot, try reseating your video card or trying a spare if you have one. -- Roger Knopf <standard disclaimer applies> SCO Consulting Services "The True Believers will...formulate uunet!sco!rogerk or rogerk@sco.com a message that even a monkey could 408-425-7222 (voice) 408-458-4227 (fax) understand." --Jeff Tye
rogerk@sco.COM (Roger Knopf 5502) (07/17/90)
In article <9Z1yL1w161w@cdthq> gary@cdthq (gary) writes: >max@mthvax.cs.miami.edu (Max Southall) writes: >> >/dev/tty01: getty keeps dying - there may be a problem >> Yup, see it all the time, randomly. > >Check and make sure all the handshaking lines on your serial >port are terminated (either looped back to it's opposite number >on the back of the computer, or driven by your terminal). Noise >on open lines has caused me problems in the past. A very helpful suggestion if this were a serial line, but this is a console multiscreen login (tty01-tty12 are console). -- Roger Knopf <standard disclaimer applies> SCO Consulting Services "The True Believers will...formulate uunet!sco!rogerk or rogerk@sco.com a message that even a monkey could 408-425-7222 (voice) 408-458-4227 (fax) understand." --Jeff Tye
sean@utoday.UUCP (Sean Fulton) (07/21/90)
Forgive me if this seems innane, but I believe you can't specify "respawn" in /etc/inittab or else exactly what you describe will happen. I remember this happened a while back with our Xenix 2.3.2, the docs said put it at respawn to get it to come back on logout, but the real solution seems to be enable the port (affecting /etc/ttys) but leave "off" in the inittab field. -- Sean Fulton *The above does not reflect the opinions of my employer* uunet!utoday!sean *If you don't believe me, ask him* Unix Today! (516) 562-5430 =>To subscribe, mail your request to: uunet!utoday!circ<=