The Unix Cluster FAQ

The information contained on the University of Virginia’s Department of Information Technology and Communication (ITC) website is provided as a public service with the understanding that ITC makes no representations or warranties, either expressed or implied, concerning the accuracy, completeness, reliability or suitability of the information, including warrantees of title, non-infringement of copyright or patent rights of others. These pages are expected to represent the University of Virginia community and the State of Virginia in a professional manner in accordance with the University of Virginia’s Computing Policies.

Frequently Asked Questions

This file contains answers to general questions concerning the Unix Cluster (blue.unix.virginia.edu).

Q6. I was running a program on blue.unix and the system killed the job. Why?

Q7. How can I use the unix "find" command to locate my files within the .snapshot directory?

Q8. I had a personal web page before the move to the cluster. How do people access my web pages? Do I need to change my web page?

Q9. I have more than one Unix account. How will the blue.unix cluster affect my mail?

Q10. What issues could arise while using blue.unix with programs such as cron, at, and calendar?

Q11. I would like to set extended permissions so that certain other people can read from and write to certain files. How can I set those permissions?

Q16. I purchased a disk wedge on faraday/watt and cannot find my snapshots. Why?

Q17. I got an NFS write error on host home1-2.Virginia.EDU. What does that mean?

==========================================================================
Q1. What are the advantages of blue.unix.virginia.edu?

A1. The Unix Cluster, named blue.unix.virginia.edu, provides a multi-machine environment where you gain access to your home directory from any of the nodes in the Cluster. If the node you are working on happens to go down, you can simply connect to the Cluster again which will automatically log you into another node to continue your work. If additional nodes are needed, then this new machine configuration makes it much easier to add new hardware to improve performance. All but a few users' files have been moved from their old RS6000 accounts to the new Network Appliance (NetApp) file server, which provides quicker response when loading and updating files. For more information about the migration to the Cluster and the NetApp, see the document Migration to the NetApp

--------------------------------------------------------------------------
Q2. How do I log into the Cluster?

A2. You can use telnet to connect to blue.unix.virginia.edu. For eXceed users who use the chooser to select a system, choose a cluster node name, such as node1.unix.virginia.edu, node2.unix.virginia.edu, etc.

Users of the defunct stand-alone RS6000's use the same ID and password they those machines. All files that were on the old system will be available. The cluster is comprised of several nodes. When you log in, you will be assigned the least loaded of the nodes. Your mailbox and files will be available to you from any node.

--------------------------------------------------------------------------
Q3. How can I restore files that I've deleted from my account?

A3. ITC provides a utility called "snapshot" for people who have their home directories on the NetApp server. Most people with accounts on blue.unix have their home directories on the NetApp. This utility can be run from the Unix prompt by typing the command snapshot or by choosing the Snapshot option from the Cluster Commands menu in Umenu. For information about using snapshot, see the document: Umenu Reference Manual For a more detailed description of snapshot files on the NetApp, see the document: Retrieving Deleted Files

--------------------------------------------------------------------------
Q4. What will be my account's quota on blue.unix?

A4. Quotas: Disk space quotas for home directories on the NetApp are different from the quotas on the other RS6000s in a couple of ways.

If the "quota" command tells you that you have no quota, then please report the problem to consult@virginia.edu. If you log into another node, the quota command will work properly.

--------------------------------------------------------------------------
Q5. If I compile a program on blue.unix, will it run on other RS6000s?

A5. The nodes which comprise blue.unix are running AIX 4.2.1 as their operating system. Executing a program which has been compiled on blue.unix should work on another RS6000 with that same OS version. If you executed such a program on an RS6000 with a lower OS version, the program would not work.

--------------------------------------------------------------------------
Q6. I was running a program on blue.unix and the system killed the job. Why?

A6. Blue.unix is a group of general use Unix machines. Because of the wide range of applications that these machines will handle, ITC does not allow users to run compute jobs that last more than one hour on these machines. ITC does, however, allow these types of jobs on the SP2 platform.

--------------------------------------------------------------------------
Q7. How can I use the unix "find" command to locate my files within the .snapshot directory?

A7. The home directory server does not support the "find" command inside of the .snapshot directory. Since it does support the "ls" command and it supports "find" within .snapshot's subdirectories, you can use a shell script to combine the "find" and the "ls" command to look for your files. A working script is:

 #!/bin/ksh for i 
  in `"ls"` do cd $i echo searching `pwd` find . -name filename -print 
  cd .. done

Where filename is the name of the file.

--------------------------------------------------------------------------
Q8. I had a personal web page before the move to the cluster. How do people access my web pages? Do I need to change my web page?

A8. The home directory server stores your personal web pages as well as the other files for your Unix account (such as your startup files and your mail folders). The address that ITC recommends that you give out to people for your personal web pages is: http://www.people.virginia.edu/~Your_Computing_ID where Your_Computing_ID is the combination of your initials, letters, and numbers that you use to log into ITC's Unix machines. If you use this URL, others will see your web pages just as they did before. The only change that is recommended is to edit your existing pages to use the new server name. NOTE TO ITC STAFF: ITCWeb pages are still on the local hard disks of the machine jm.acs.virginia.edu. You must do any editing to those pages on jm.

--------------------------------------------------------------------------
Q9. I have more than one Unix account. How will the blue.unix cluster affect my mail?

A9. If you have accounts on more than one Unix machine that uses the home directory service (accounts on "blue," holmes, and jm), you have the potential for your mail to be "scattered." To keep all your mail going to one system, create a .forward file that has the full address for the system on which you read your mail, such as: Your_Computing_ID@unix.mail.virginia.edu DO NOT use: Your_Computing_ID@virginia.edu as you may cause the system to develop a "mail loop." You may either create this .forward file with a Unix editor or via the "System Customization"->"Mail Preferences"->"Forward All Mail to Another Address" options on the umenu. If you have questions about forwarding your mail, please send them to "unix-consult@virginia.edu" or call the ITC Help Desk at 924-3731.

--------------------------------------------------------------------------
Q10. What issues could arise while using blue.unix with programs such as cron, at, and calendar?

A10. Cron (and crontab) and at all put their files on the local file systems of the machine on which you issued the commands. If you use these programs, please make sure that you use same node of the cluster (i.e. node1, node2, etc.) whenever using these programs. Otherwise, you have their job files scattered among several different machines, which could damage your work as well as adversely affect the system when several machines try to issue the same commands simultaneously. The calendar program, however, reads a file named calendar into your home directory. If you use this program, you may receive more than one message if you have accounts on more than one machine. You will get a calendar message from each machine that can read your calendar file. If you set permissions so that only you can read the file, only a single message will be sent from www.people.virginia.edu. The command to issue to restrict access to your calendar is (from the Unix prompt):
chmod 600 calendar

--------------------------------------------------------------------------
Q11. I would like to set extended permissions so that certain other people can read from and write to certain files. How can I set those permissions?

A11. If you want others to be able to read from and write to your files, please write to "consult@virginia.edu" and ask for a Unix "group." With a "group," you can define a certain set of users that you want to have certain levels of access. ITC must create and change the membership of a group, so include a list of members with your request. To give a group certain access to a file, type: chgrp groupname filename where groupname is the name of the group ITC defines. Then type: chmod g+r filename -- for "read" permission chmod g+w filename -- for "write" permission chmod g+x filename -- for "execute" permission You may combine these commands (i.e. chmod g+rx) to establish a combination of permissions. The new cluster does not support the AIX operating system's access control lists functions (acl's). For information on how to use acl's on other AIX machines, please read Appendix C of document U002.

--------------------------------------------------------------------------
Q12. How do I use the "talk" program to talk to others on blue.unix?

A12. The following information is included in the talk.doc at http://www.itc.virginia.edu/desktop/unix/docs/talk.doc.html For those wanting to talk to another on the Unix Cluster: 1. From the main menu of Umenu, select "Cluster Commands." 2. Select "Gather," 3. Select "Friends." 4. Enter the ID of the person. If the person is using the Unix Cluster, the command will show information similar to: node4: mst3k pts/3 Feb 2 07:34 (bogus.domain.Vir) The first bit of information shows which "node" of the cluster the other person is using. In this case, it is node number 4. You will need this information later. 5. Press the return key to return to the "Gather" menu. 6. Type "m" then press enter to return to the main menu. 7. Select "Go to Unix" from the main menu. 8. To use the talk command, enter the command: talk ID@node#.unix.virginia.edu where ID is the computing ID of the person to whom you want to talk and node# is the node that person is using. For example, talk mst3k@node4.unix.virginia.edu would be the command to talk to the user in the example above. The rest of the instructions are the same as described in http://www.itc.virginia.edu/desktop/unix/docs/talk.doc.html.

--------------------------------------------------------------------------
Q13. How should I invoke "screen" on the cluster?

A13. When using the "screen" program, make sure that you note which node of the cluster that you are using. When you invoke "screen," that node keeps your session to itself and not share with the other nodes. To start "screen," type: screen -R This command will reattach you to the first screen session that it finds. If successful, it will ignore all other command line parameters. If it cannot find a previous session, it will start a new screen session, as if you did not type the "-R."

--------------------------------------------------------------------------
Q14. How can I write a .rhosts file for the cluster?

A14. ITC does not recommend that users use a .rhosts file. Using this file can create a security hole that others can use to gain access to your Unix account. If you need to create one for the cluster, you will need to put the name of each node of the cluster. For example, user mst3k would have a file: node1.unix.virginia.edu mst3k node2.unix.virginia.edu mst3k and so on. If ITC adds new nodes to the cluster, you will have to add an entry for those new nodes. You can see how many nodes the cluster has by typing "gather la" at the Unix prompt to see how many machines reply. Using: blue.unix.virginia.edu mst3k will not work, since you must use the real names of the machines with this type of connection.

--------------------------------------------------------------------------
Q15. Why can't I run GCG on the cluster?

A15. The Genetics Computer Group, the makers of the GCG program, announced that they will no longer support GCG under AIX, the operating system on the RS/6000's. GCG users with accounts on avery.med.virginia.edu may use the machine crick.med.virginia.edu for their work. Like the Unix cluster, these machines share a common loginid/password database, so you may use your avery password to access that system. Other GCG users should request an account on ITC's Unixlab machines. GCG will be available on the machine named sgi-crimson.unixlab.virginia.edu for those users.

--------------------------------------------------------------------------
Q16. I purchased a disk wedge and cannot find my snapshots. Why?

A16. ITC is mounting personal disk wedges to the second Network Appliance, home2.virginia.edu. If you are unable to access your disk wedge, write to "consult@virginia.edu"

--------------------------------------------------------------------------
Q17. I got an NFS write error on host home1-2.Virginia.EDU. What does that mean?

A17. When you reach your quota on your blue.unix account, you will see this error message on your screen. Until you delete or compress some of your files inside of your home directory, you will not be able to create or modify the files inside your home directory. You will see this error each time a program tries to write to your home directory, including many mail programs that keep some temporary files inside your home directory.

--------------------------------------------------------------------------
Q18. Why can't I cancel the print job I just submitted through lpr?

A18. You can only cancel print jobs from the node of the cluster on which you submitted the print job. If, for example, you submit a print job on node6.unix, you will be unable to cancel that job on the other cluster nodes. You can, however, determine which node you used from the output of the lprm command, whether or not the command succeeded in removing the job from the print queue. For example, if you submitted a print job from node10 and then ran the lprm command, you would get output similar to: print.itc.Virginia.EDU: dfA002node10.unix.Virginia.EDU ^^^^^^^^^^^^^^^^^^^^^^^^ As you can see, the original sending host is kept so you can connect that specific node and then run the lprm command successfully to remove the print job.

Summary of Unix Machine Names
Cluster	blue.unix.virginia.edu
Nodes	node1.unix.virginia.edu, node2.unix, etc.
Web Server	www.people.virginia.edu
Receiving Mail	unix.mail.virginia.edu
Sending Mail	smtp.mail.virginia.edu

The e-mail configuration on blue.unix is slightly different from that which you will see on a single, stand-alone server. Mail handling for blue.unix is done by two machines which are separate from the nodes of the cluster. The machine which delivers incoming mail is called unix.mail.virginia.edu, and the machine which handles outgoing mail is called smtp.mail.virginia.edu. Mail inboxes for people using blue.unix reside on unix.mail. These names were chosen for several reasons. For people on the cluster who wish to use Eudora or Netscape to read their e-mail, these names aid in configuring those mail reading programs. For example, Netscape's mail configuration asks for the name of your "Outgoing Mail/SMTP Server:". The name of this server for people using blue.unix is smtp.mail.virginia.edu.

What advantage does this configuration have? If blue.unix was only one machine, and it happened to crash, then you would not be able to log in or otherwise get to your e-mail. Since blue.unix consists of several nodes, if the node you are on crashes, you can log onto another node and still be able to receive, read, and send e-mail.

For:

Help

About

Standards & Policy

Frequently Asked Questions