Adding New Users

• Fill out a form and have it signed by the faculty supervisor or graduate/undergraduate co-ordinator
  
• Login as a superuser into Neptune using the command su.
  
• Look into the password file to get the last user id#. The user id of the new user will be the next #. The password file is in /etc/yp/passwd
  
• Goto the directory /local/adm/system/accounts and type adduser. Set up the account and ask the user to enter the password he/she wants to have provided it is 6 to 8 characters long and difficult to guess. It is important to point out not to pick any word in any language dictionary as it is very easy to crack them.
  
• After creating the account setup the alias. Add the new user to any groups. The files to be accessed are /etc/aliases and /etc/yp/group. Make sure his entry into the graduate or undergraduate list is done. After the files are edited, propagate the changes to the yellow page clients by doing the following
  #cd /var/yp
  #make
  
• Allocate the quota on the home filesystem of the user. The steps for the allocation are:
  #login on the home directory machine
  #edquota username
  Undergraduates start off with 8-10 Megabytes. Graduate students around 15 Megabytes. Post-docs and faculty allocations change depending on group. Any large request need to identify a research project and may require contribution towards purchasing more disk drive space.

Deleting Accounts

Periodically, it is necessary to delete the accounts of users who no longer need access to the physics computer system.

• The file nuke.next in the directory /local/adm/system/accounts on neptune contains information regarding the next batch of accounts to delete. The file nuke.mail contains the e-mail messages sent from users whose name was on the previous nuke list. This should be consulted when determining if a user's account should be deleted.
• Warn the people whose accounts are on this list that their accounts are going to be deleted and give them ten days to clean out their accounts. After that time, send out one more five day warning to the people who have not yet responded to the first warning.
• Delete the user's directory. Also remove the appropriate entry in /etc/yp/passwd. Remember to remove all entries in /etc/yp/group, and make any necessary changes to /etc/aliases, either by deleting the record or setting up a forwarding address. Go to /var/yp and do a make.
• Make all appropriate changes in the nuke files for documentation.

User Backups

Backups are performed via pre-written scripts which are automatically scheduled using cron. We use the filesystem based backup mechanisms. The commands we use are dump and restore on the Alphas and ufsdump and ufsrestore on the Suns. The scripts are called /.bkup0.userdata and /.bkup2.userdata on neptune and /.bkup0.userdata and /.bkup2.userdata on quark.

Level 0 (Full) Backups

This is a complete backup of all user filesystems and is done every week on Friday. It is done on neptune at 3:30 AM every Saturday morning using the /.bkup0.userdata. It dumps /phys/jupiter0 from jupiter, /moffice from mildew, /var/mail, /phys/neptune0-3, /phys/cmt locally to the DLT tape. A second level 0 dump starts on quark at 2:30 AM Saturday morning using the /.bkup0.user. It backs up /users from wombat, /heeusers from hep01, /web and /quark/users locally to the DLT tape drive. Each DLT tape can hold about 10 GB of compressed data. At the current filesystem sizes, the tapes need to be changed every 3 weeks on both neptune and quark. Since the backups happen on Friday night during the early morning hours of Saturday, you need to have the tapes changed by Friday evening before leaving work. The Level 0 backup tapes are kept for several years.

Level 2 (Incremental) Backups

This is an incremental backup of all files modified on the filesystem since the last level 0 backup. This is done nightly excepting on the night Level 0s happen. On neptune we use the 8 MM tape drive with about 1.2 GB capacity and on quark we use the 4 MM DAT tape with 1.0 GB capacity. A tape change is needed every Friday. Four weeks worth of backups are kept and the tapes are reused in cycles. We have four tapes labeled Incremental Backup A, B, C and D near the tape drives on neptune and quark. Replace the tapes in alphabetical sequence.

Restores

Before beginning the restore procedure, determine a directory which has enough free space. The tape can be restored in that directory. We generally pick /var/tmp. You will also need to figure out which of the tapes we need to restore the file from. It could be from the level 0 backup tape if the file was around on Friday night of last week. If the file was modified since Friday night's backup, it will be in the level 2 tapes. All restores are done on maxwell, as it has extra tape drives. Use the DLT 7000 to restore DLT's dumped to by hee or maxwell. Use the internal 4mm drive to read from quark's 4mm tape. The DLT is /dev/rmt/3cn and the 4mm tape is /dev/rmt/2h.

1. Login as root or su on quark or neptune
2. Create a directory restore in the /var/tmp directory.
3. Change to the directory by typing cd /var/tmp/restore.
4. Position the tape on the correct file to restore from. You may need to use

5. mt -f rewindtapedevice rewind (e.g /dev/rmt/3cn, /dev/rmt/2h)
   mt -f rewindtapedevice offline
   mt -f nonrewindtapedevice asf # (this rewinds and then forwards # records)
   

6. ufsrestore -if nonrewindtapedevice
   

7. You can do ls, cd, add subdirectory or file, and type extract
8. The system will say

   You have not read any tapes yet.
   Unless you know which volume your file(s) are on you should start
   with the last volume and work towards the first.
   Specify next volume #:
   

9. You type 1 and hit return. The files you requested will be extracted. The system will then ask to set the owner mode and protection. You can say yes for that.
10. quit the restore.

Positioning the Tapes

Solaris:
mt -f Non-RewindDevice eom

Digital Unix:
mt -f Non-RewindDevice seod

mt -f Non-RewindDevice fsf 50