To Power or Bring Down UNIX __ _____ __ _____ ____ ____ 1) Log in and execute the command "reboot". DO NOT FAIL TO DO THIS IN ANY CIRCUMSTANCES WHATSOEVER. 2) If the power switch on the front panel of the 11/45 is in the "LOCK" position, turn it to the "POWER" position. Lower the "ENABLE/HALT" switch on the 11/45 to "HALT". 3) Turn off the disk drives. On the RK's, this entails switching the LOAD/RUN switch to load; for the DIVA's, flip the right hand switches down. DO NOT TURN THE POWER OFF WITH DRIVES IN THE "READY" STATE! MAKE SURE THAT THEY ARE ALL OFF!!! 4) Turn the key on the 11/45 front panel to "OFF". Reach inside the DIVA's behind the swing out card cage and turn off the two power supplies. This will turn off power to most everything, but power will still be sent to the DIVA's, and in the 11 power supply. There are parts of the power distribution box in the base that are always live. To Boot UNIX __ ____ ____ 1) Lower the "HALT/ENABLE" switch to "HALT" on the 11/45 front panel. 2) If the system was powered down, turn on the power on the 11/45 front panel. If the front panel is locked (left middle cylinder is in lock position), unlock it by turning to the power position. Set the right top switch on the front panel to con phys and the switch below to display register. 3) Make sure that there is a system console, that it is turned on and that it is set to 9600/300 baud rate. 4) Make sure the disk pack in RK drive 0 is the correct one. It should be labelled something like "HDev UNIX" or "SDev UNIX". Set the "LOAD/RUN" on RK drive 0 to "RUN". Make sure that the pack in RK drive 1 is labelled "SWAP". Load it into drive 1 if it's not there already, and hit "RUN". 5) Make sure that the correct DIVA packs are in the correct drives. The pack in the top drive should be pack #1, labelled "/root & /src". The correct pack for the bottom drive is #2, labelled "/usr". If the DIVA's were powered down, turn them on by doing the following things: There are two switches inside the DIVA's on their power supplies that must be switched on - the switches are behind the swing out card rack at the back of the disk cabinet. Also flip the right hand switch on the diva front panel (POWER ON/OFF) up (to ON) on both drives (wait until one drive is spinning before turning on the other to prevent excessive surge current), and wait for the green "READY" lights to come on. 6) Set the disk bootstrap address into the "ADDRESS/DATA" switches on the 11/45 front panel. This is currently 773220. 7) Press "LOAD ADDRESS" on the 11/45 front panel. 8) Raise the "HALT/ENABLE" switch to "ENABLE". 9) Press "START". A "@" will be printed on the system console. 10) Enter "773030" into the "ADDRESS/DATA switches. (This is to bring up the system with as few things going on as possible for salvaging.) 11) Load the system by typing the name of the current system file, which is always shown on the computer card in the holder on the front of the 11, and hit a when done. If you make a mistake, don't worry. Just type and the loader will give you another try. The system should now load. (Disks frob, etc.) 12) After a bit, UNIX will type out how much memory it thinks the system has. 13) When you get a prompt (">"), salvage the disk(s). You can either do this by explicity invoking the right commands, or you may use a command file that automatically does the right thing. To use the command file: Type ". idd" To explicity salvage things: Execute the command "icheck /dev/rdv1 /dev/rdv10 /dev/rdv0". Execute the command "dcheck /dev/rdv1 /dev/rdv10 /dev/rdv0". From the first phase (icheck) you should get a list of the number of: special files (devices), regular files, large files, huge files, directories, indirect blocks, blocks used, and free blocks. Anything else is an error. "Missing" errors can be ignored and patched later. Any other error is fatal. (Go to Trouble/Expert.) From the second phase, if there are no errors, all you will get is a line containing the file system device. If there are any errors, you will get a list of I-node numbers, and two columns of numbers of "entries" and "link cnt". If any of the I-nodes listed have more entries than their link count, the disk will have to be repaired. (Go to Trouble/Expert.) If both the "entries" and "link count" are zero, or the link count is greater than the number of entries, you can ignore it and it can be patched later. (See Trouble/Expert.) 14) Type "ts" on the system tty to bring the system up in multi-user mode. You might be asked to set the date/time manually. The format for entering the date/time is mmddhhmm where each pair of digits is a 2-digit decimal number representing the month, day, hour (24 hour) and minute. The system should now begin to run in normal mode, and will type "login:" on all connected terminals. EXAMPLE: @xdunix mem = 2700 >. idd /dev/rdv1: missing 5 <-- missing blocks are ok spcl 19 files 442 large 56 direc 19 indir 68 used 3432 free 23464 /dev/rdv10: spcl 0 files 872 large 123 direc 45 indir 102 used 7234 free 53464 /dev/rdv0: spcl 0 files 769 large 102 direc 34 indir 124 used 9673 free 50174 /dev/rdv1: entries link cnt 3077 0 1 <-- High link count is ok 3200 0 0 <-- Zero for both is ok 3186 1 0 <-- Low link count is bad /dev/rdv10: /dev/rdv0: >ts Unable to set time. SET TIME MANUALLY: 05251730 (for May 25 5:30 pm) login: Trouble/Expert ______________ This generally means Noel, Liza, or Larry. If we're not around, leave a description of the problem or need somewhere (like ARPA mail - JNC@MIT-XX) and notes in a few obvious places. If we're around, leave things exactly as they are and go get one of us.(If the system looks like it's doing something, and not what it should, hit "HALT" before leaving.) Otherwise, you may either want to leave everything as is (but put "HALT" down, so it won't go berserk while you're not around) and leave a message, or attempt to shut things down. If the problem looks like software, you might try saving some information for me. If the system stopped of it's own accord, note the PC (displayed in the ADDRESS lights.) If it typed anything on the console ("trap type " , etc) record it. Try and take a dump by using the following algorithm. 1) Stop the system (i.e. "HALT" down.) 2) Make sure the SWAP pack is in RK1. 3) Enter 44(octal) into the "ADDRESS/DATA" switches, and hit "LOAD ADDR". Hit "START". The RK drive should blink a bit, and then stop, and the system should stop with all the "DATA" lights on. If it did, you won. RK 1 now contains a dump of the crashed system. From there, you can either shut the system down (start at step 2) or reload if you feel brave. An exception to the above rule is when the system prints out a message of the form no room on dev 0/0 which means that you just ran out of room, or err on dev 0/0 bn1234 er2 4702 err on dev 8/2 bn12345 er100320 15123 or some such where the bn number and the last number may change. "er2" means that the disk drive detected a checksum error when reading/writing a block. The operation is automatically retried and will usually succeed. If you get an error five times in quick succession from the same block, the system couldn't read it at all. This is fairly serious; if it happens during salvaging the disks the system should not be brought up. If this starts to happen a lot, or if you get an error of any other form, you probably want to stop instantly and save your work as the system may crash fatally at any instant, and may have done so already. In any case, it generally indicates a serious error that should be looked at.