Paul Humphreys rambles on....
News and Views

20050324 Thursday March 24, 2005

More on the lab; the console server

For a remote lab to work - and I define a remote lab where the engineers can be separated from the lab by the distance of say two floors or in a building or thousands of miles, remote manipulation of the machine in terms of reboots etc is vital.

Chris took the console software that is in the public domain and tailored it for our own use. We are planning to work out how to put our version in the public domain or merge our changes in with the one out there still.

The idea is you start with terminal devices either locally connected like the ones we use or network devices which usually support the telnet protocol to make contact with the ports on them. In fact Sun's high end servers like the F6900 and F4900 have system controllers that also support telnet protocol which is handy. We then take a standard Sun server in our case a V210 load Solaris on it give it a bit of redundancy with disk mirroring and then load Chris's console software. The console software has a 'stub' config file that tells it to go and talk to our Javalabtool booking database which has all our console entries. The console software also has a 'stash' file to use if this connection fails initially. The software then grabs all the console entries for the database that pertain to itself. It then makes the connection or opens the terminal device and presents these on port 5000 of the console server. The console server also then writes the various ports it has setup into the NIS+ table serialports that we use to store ALL our console entries worldwide. Chris has written a 'console' command that given an argument of a lab box will then search the NIS+ table, find the entry and make contact to the appropriate server and port using telnet. I have simplified this description slightly.

The sorts of things the software does for us:

Automatically maintains a log file of all console transactions for later use per console port configured.

As you make connection to the port the console command warns you if the machine is booked and to who.

Maintains one console writer and numerous console readers for multiple access to a console port. A known escape sequence flips you from being a reader to a writer - an event that is logged in the console log

Demands a password before connection is made configurable per host so different lab boxes can have different passwords (stored in javalabtool)

Support the sending of various escape sequences to send a break to a host stop/start logging into the log file etc

These are the basics. It does a whole heap more. But it is a core competancy of a shared lab to be able to do this sort of stuff. When you reboot or send a break to a machine in Singapore thousands of miles away it blows your mind.

( Mar 24 2005, 03:57:20 AM PST ) Permalink Comments [0]

Trackback URL: http://blogs.sun.com/paulhu/entry/more_on_the_lab_the
Comments:

Post a Comment:

Name:
E-Mail:
URL:

Your Comment:

HTML Syntax: NOT allowed

Archives
Language
Links
Referrers