hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) (01/12/90)
uVaxII, Ultrix3.1 We just had to shutdown and reboot a system to switch some disks. We had to halt the thing during one of the reboots. Ran fsck on /dev/ra0a after booting up single user - it made some minor repairs on one inode. After that the logins to the hardware ttys aren't there even though the gettys are running. Additionally the login prompt nolonger displays the hostname: just 'login' The network ports and the console and X term logins all work but the dhv ports have gone south. I checked /etc/ttys /dev, - can't see any thing wrong - a 'cat /etc/ttys >/dev/tty08' gets nothing out tty08... Our dhv is a dilog 1620 16 port board. Uerf doesn't report any errors... I put a new kernal on and rebooted - no fix. suggestions? hurf -- Hurf Sheldon Network: hurf@ionvax.tn.cornell.edu Lab of Plasma Studies Bitnet: hurf@CRNLION 369 Upson Hall, Cornell University, Ithaca, N.Y. 14853 ph:607 255 7267 "And the walls came tumbling down"
grr@cbmvax.commodore.com (George Robbins) (01/12/90)
In article <9527@batcomputer.tn.cornell.edu> hurf@tcgould.tn.cornell.edu (Hurf Sheldon) writes: > uVaxII, Ultrix3.1 > > We just had to shutdown and reboot a system to switch some disks. > We had to halt the thing during one of the reboots. Ran fsck on /dev/ra0a after booting up single user - it made some minor repairs on one inode. > After that the > logins to the hardware ttys aren't there even though the gettys are running. > Additionally the login prompt nolonger displays the hostname: just 'login' > > The network ports and the console and X term logins all work but the dhv ports > have gone south. I checked /etc/ttys /dev, - can't see any thing wrong - > a 'cat /etc/ttys >/dev/tty08' gets nothing out tty08... > Our dhv is a dilog 1620 16 port board. Pretty spooky! Check that the /dev entries still show up as "character" type special files with rational major and minor device numbers, and also reasonable ownership/ protections. Check that /etc/ttys and /etc/gettytab haven't been "improved". Look at /usr/spool/mqueue/syslog (or wherever you have syslog going) - this is where init/getty/login report their problems. Finally if you have hardcopy take the inode number and use "dcheck -i" to see what it was. See if anything interesting has shown up in lost+found. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@uunet.uu.net Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) (01/12/90)
Posting my own followup: I found that /etc/gettytab had been rewritten with the top 3690 bytes of /etc/passwd - this is the second time under Ultrix3.1 on 2 different systems this has happened - ????? hurf -- Hurf Sheldon Network: hurf@ionvax.tn.cornell.edu Lab of Plasma Studies Bitnet: hurf@CRNLION 369 Upson Hall, Cornell University, Ithaca, N.Y. 14853 ph:607 255 7267 "And the walls came tumbling down"
envbvs@epb2.lbl.gov (Brian V. Smith) (01/12/90)
In article <9533@batcomputer.tn.cornell.edu>, hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) writes: < < Posting my own followup: < I found that /etc/gettytab had been < rewritten with the top 3690 bytes of /etc/passwd - this < is the second time under Ultrix3.1 on 2 different systems < this has happened - ????? < We had the same problem in Ultrix 3.0. At first I thought I had rcp'd the file on top of itself which is guaranteed to clobber the file in a similar way, but then I realized that I hadn't changed /etc/gettytab (ever), so I hadn't run rcp at all on it. It happened on two of our six machines. Any ideas? ____________________________________ Brian V. Smith (bvsmith@lbl.gov) Lawrence Berkeley Laboratory I don't speak for LBL, these non-opinions are all mine.
steven@pacific.csl.uiuc.edu (Steven Parkes) (01/12/90)
In article <4614@helios.ee.lbl.gov>, envbvs@epb2.lbl.gov (Brian V. Smith) writes: > From: envbvs@epb2.lbl.gov (Brian V. Smith) > Newsgroups: comp.unix.ultrix > Subject: Re: SYSTEM logins weird > > < I found that /etc/gettytab had been > < rewritten .... > We had the same problem in Ultrix 3.0. ... We've had a simillar problem on 3500's under 3.1 ... it wasn't passwd but part of /usr/spool/mail/*. Sounds like more than a fluke ...
grr@cbmvax.commodore.com (George Robbins) (01/12/90)
In article <4614@helios.ee.lbl.gov> envbvs@epb2.lbl.gov (Brian V. Smith) writes: > In article <9533@batcomputer.tn.cornell.edu>, > hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) writes: > < > < I found that /etc/gettytab had been > < rewritten with the top 3690 bytes of /etc/passwd - this > < is the second time under Ultrix3.1 on 2 different systems > < this has happened - ????? > > We had the same problem in Ultrix 3.0. > At first I thought I had rcp'd the file on top of itself which is guaranteed > to clobber the file in a similar way, but then I realized that > I hadn't changed /etc/gettytab (ever), so I hadn't run rcp at all on it. > > Any ideas? Remember that the startup script edits the /etc/gettytab and /etc/ttys files to update the system name. I suspect this kind of thing might be happening if your system croaks without haveing written out the new version. On the other hand, unless it died withing a few moments after startup, it *should* have written the stuff from the buffer pool... -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@uunet.uu.net Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
greg@duke.cs.unlv.edu (Greg Wohletz) (01/13/90)
In article <9533@batcomputer.tn.cornell.edu>, hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) writes: < < Posting my own followup: < I found that /etc/gettytab had been < rewritten with the top 3690 bytes of /etc/passwd - this < is the second time under Ultrix3.1 on 2 different systems < this has happened - ????? < hmm, this has happened several times to me as well. Ultrix 3.1 has been quite a headache ever since I upgraded from 2.0. I've posted two messages about various problems I have had and recieved NO replies whatsoever. Doesn't anyone from the ultrix group read this newsgroup? --Greg
avolio@decuac.dec.com (Frederick M. Avolio) (01/13/90)
In article <1453@jimi.cs.unlv.edu> greg@duke.cs.unlv.edu (Greg Wohletz) writes: >hmm, this has happened several times to me as well. Ultrix 3.1 has >been quite a headache ever since I upgraded from 2.0. I've posted two >messages about various problems I have had and recieved NO replies >whatsoever. Doesn't anyone from the ultrix group read this newsgroup? > > --Greg Well, send me your two problems and/or repost them. Yes, people from the ULTRIX group read this (and no, I am not in the ULTRIX group) but the best bet to get fixes is still through buying support. I guess that sounds sort of capitalistic, eh? :-). The problems Hurf is having sounds like file system problems to me... But I am hard-pressed to blame it on software since very few systems (I now know of two) have had these problems. Fred
pat@orac.pgh.pa.us (Pat Barron) (01/13/90)
In article <9533@batcomputer.tn.cornell.edu> hurf@tcgould.tn.cornell.edu (Hurf Sheldon) writes: > Posting my own followup: > I found that /etc/gettytab had been > rewritten with the top 3690 bytes of /etc/passwd - this > is the second time under Ultrix3.1 on 2 different systems > this has happened - ????? I've seen this a lot, too. I've always suspected that it had something to do with the fact that /etc/gettytab is edited in /etc/rc or rc.local (to put the current system version string into the default login banner) but have never been able to prove anything.... --Pat. -- Pat Barron Internet: pat@orac.pgh.pa.us - or - orac!pat@gateway.sei.cmu.edu UUCP: ...!uunet!apexepa!sei!orac!pat - or - ...!pitt!darth!orac!pat
greg@duke.cs.unlv.edu (Greg Wohletz) (01/14/90)
In article <2877@decuac.DEC.COM>, avolio@decuac.dec.com (Frederick M. Avolio) writes: > > Well, send me your two problems and/or repost them. Yes, people from > the ULTRIX group read this (and no, I am not in the ULTRIX group) but > the best bet to get fixes is still through buying support. I guess > that sounds sort of capitalistic, eh? :-). Sorry for the flame, I've had a bad week... Anyway. We actually tried to order source code maintenance with our tape, but somehow the order got screwed up. Here is my last message again. In article <1444@jimi.cs.unlv.edu>, I write: > We have three microvax II's that we use as fileservers. Each has 3 > Wren V's and an Exabyte hooked into a Sigma scsi controller (it > emulates a uda and tms controller). They also have a Dec uda > controller hooked to an rd52 and two rx50's (yes we've had these > machines for a while...) on them. We have been running with this > configuration under Ultrix 2.0 without many problems (well a few nfs > bugs, but nothing major). Recently we got Ultrix 3.1. I installed it > on one of our microvax's and everything seemed to be going fine, I > could use the disks, and read from the Exabyte. However, when I tried > to dump the root filesystem to the Exabyte I got a write error, then > some message like ``mscp resynching controller uq2'' at that point the > system locked up. Well, I've investigated the situation a bit further, and I have discovered that (surprise, surprise) one difference between 2.0 and 3.1 is that all of the disk and tape drive stuff appears to have be re-written. Now everything (except for non-uda type drives and non-tmscp tapes) goes through this new mscp code (or at least that is what it looks like to me). Anyway looking at the code didn't reveal anything obvious, however I have noticed that I can't dump to the trusty (?) old rx50's. The first volume of the dump works fine, but if you so much as open the door to the floppy when dump asks you to insert the next volume all subsequent attempts to write to the floppy will fail (if you leave the same floppy in (without opening the drive door) for ALL of the volumes it will work...). I suspect that this problem is related to the same bug. I think at this point I'm almost convinced that it is a software bug, and not a problem with the Sigma controller, but I could be wrong. So the question is, will someone from DEC tell me if there is a known/fixable bug in 3.1 that would cause this behavior? Would if be possible to graft in the old tmscp code from 2.0 without an inordinate amount of pain? ANY information would be greatly appreciated. --Greg greg@unlv.edu <@relay.cs.net:greg@unlv.edu>
brw@hertz.njit.edu (Brian White) (01/15/90)
In article <2877@decuac.DEC.COM> avolio@decuac.dec.com writes: > >Well, send me your two problems and/or repost them. Yes, people from >the ULTRIX group read this (and no, I am not in the ULTRIX group) but >the best bet to get fixes is still through buying support. I guess We have it; I'm not convinced it's worthwhile, though... >that sounds sort of capitalistic, eh? :-). > >The problems Hurf is having sounds like file system problems to me... >But I am hard-pressed to blame it on software since very few systems >(I now know of two) have had these problems. > >Fred Uhhh, make that three, Fred. Our 11/785 (running 3.1) crashed and a *News article* showed up in place of gettytab. No login message, no baud rate set, and 8 lines of vi hipped me to the problem, but that sure was weird.... Brian White System Programmer New Jersey Institute of Technology brw@hertz.njit.edu Brian White System Programmer New Jersey Institute of Technology brw@hertz.njit.edu brw@njit.edu brw@jazz.njit.edu (yeah, I know, showoff....)
grr@cbmvax.commodore.com (George Robbins) (01/15/90)
In article <973@njitgw.njit.edu> brw@hertz.njit.edu (Brian White) writes: > In article <2877@decuac.DEC.COM> avolio@decuac.dec.com writes: > > > >The problems Hurf is having sounds like file system problems to me... > >But I am hard-pressed to blame it on software since very few systems > >(I now know of two) have had these problems. > > > >Fred > > Uhhh, make that three, Fred. Our 11/785 (running 3.1) crashed and a > *News article* showed up in place of gettytab. No login message, no baud rate > set, and 8 lines of vi hipped me to the problem, but that sure was weird.... I guess a few well placed "sync" commands in the /etc/rc.local file should cure this symptom, however it's real perverso that this file should not get flushed and forgotten, especially of the system doesn't crash immediately. After all, /etc/update should do the trick within 30 seconds. I haven't seen this particular disease, but I have occasionaly found completely trash files in my uucp queues (I know most uucp is trash, but these are obviously not the right kind of trash for where they show up 8-). -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@uunet.uu.net Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)