gbm@gbmatl.UUCP (gary mckenney) (05/08/89)
In article <1989May5.150438.13740@ziebmef.uucp} cks@ziebmef.UUCP (Chris Siebenmann) writes: }In article <17733@cup.portal.com> thad@cup.portal.com (Thad P Floryan) writes: } Do these actually accumulate on the UNIX-PC? I try to keep my system }up for as long as possible (the only problem I've found so far is that }/etc/wmgr accumulates memory, so you have to kill and restart it every }now and then), and after routine shutdowns and startups, fsck never }seems to find any problems. Have people observed the infamous SysV }disappearing-inodes problem on their machines? Are there other FS }problems people have encountered? I've run into the disappearing inode problem. I shutdown to let fsck run and it cleared the problem up. I do have difficulty setting my machine to single user mode. From the documentation the recommended way to do this is to type "init s". On my machine this will successfully reboot the machine eventually but never to quite make it into single user mode. Any suggestions are appreciated. gbm
sheldon@quest.UUCP (Scott S. Bertilson) (05/09/89)
Your confusion about "init s" getting you to singleuser mode is due to the fact that "/etc/profile" intercepts the shell before it gets to you...it finishes the shutdown and prints the "hit RETURN to reboot". It does a "trap '' 1 2 3" to stop you from getting out without rebooting. I've modified mine so that it does a "trap 1 2 3" just before it does the "read inputline". That will allow you to hit the DEL key to interrupt and get a shell prompt. -- Scott S. Bertilson ...uunet!rosevax!rose3!quest!sheldon scott@poincare.geom.umn.edu
shevett@labii.UUCP (Dave Shevett) (05/10/89)
In article <1762@gbmatl.UUCP> gbm@gbmatl.UUCP (Root) writes: >In article <1989May5.150438.13740@ziebmef.uucp} cks@ziebmef.UUCP (Chris Siebenmann) writes: >}In article <17733@cup.portal.com> thad@cup.portal.com (Thad P Floryan) writes: >} Do these actually accumulate on the UNIX-PC? I try to keep my system > >I've run into the disappearing inode problem. I shutdown to let fsck run >and it cleared the problem up. > I do too, but it's awfully annoying. The biggest margin I've had happened last night when I came home to find 0 (count 'em 0) I-nodes free. I promptly killed off a few hundred files and ran an Expire (running News, a BBS, 3 terminals, and other goodies on a 67 meg drive with 1.5 meg RAM, v3.5 sys). Expire ended, and I logged in again - Whoa! 18000 free blocks, but only 238 free Inodes. ("cant be right" sez I) - I reboot. Fsck runs twice, and a lot of hemmin and hawin while it does so, and a quick login shows 18800 free blocks, and, lo and behold, 6200 Inodes. I can handle a few inodes lost a day, but 6000??? That's half my table. Is there any way to recover these things without shutting down??? I kind of like having the high PID counts :-) PS - Can anyone send me their cron entry for expire? I have a shell I run (Expshell) that runs 3 expires. I've had hassles putting it in cron. I'd like it to run every other day. (normally I do a 'nohup ksh Expshell &') /----------------+ "The shortest distance +----------------------\ | Dave Shevett | between two puns is a | Labyrinth II BBS | | W. Trenton, NJ | straight line..." | shevett@labii.UUCP | \----------------+ - Doc Webster +----------------------/