[AG-TECH] Node Setup Wizard problems
Lewis Grantham
l.grantham at ucl.ac.uk
Wed Apr 14 10:17:02 CDT 2004
hi Randy,
try this route, it worked out here ok:
a. go to each machine in turn, and run the venueclient as a personal
node (VenueClient.py --personalNode). If any cert request wizards pop
up, cancel them (over and over) til they go away. you need each venue
client to run cleanly as a personal node without grumbling - that should
mean all certs and IDs are good, and proxies are fine on each unit. you
should get a local log on all three machines to let you know whats
occuring. if you need to, reimport IDs and certs from the
Venue Client > preferences > Manage Certificates menus
You can check they are all Ok there also. Only ANL issued certs and IDs
are anygood in the longterm. Importing other CAs and IDs doesnt totally
work last we tried. Take some time to check all cert parts. The
default ID, a globus CA and ANL CA cert are needed, and should be
imported in sequence (globus CA , ANL CA , then personal cert).
If you reimport - try removing all the older IDs and reimporting. I
keep a floppy to hand with both CA certs and my trusty ANL ID for this
tasks. Paranoid maybe, but a good clear out works wonders 9/10 times.
(By the by - it is very vital you export a safe, known good copy of your
ID for later use and retrieval).
when each machine runs venue client with personal node services cleanly,
you can set up the wider multimachine options. I would not try too much
further til this step works, and you can start VenueClient with personal
node services on each unit manually.
Also, dont waste time 'managing' your node yet (adding services etc)-
all those parts change later.
You should not at this stage be running any other tools manually. No
command line Node services or Ag service managers - just the
VenueClient.py with the --personalNode option. try to get away from
older AGTK thinking - theres a lot of differences, largely undocumented
(sorry Ivan, but there is ;)
b. choose a machine that your users will typically start the toolset
from. (Almost always this is the display PC). This machine will
continue to run :
VenueClient.py --personalNode
Make sure the shortcut or script file includes this option.
Never attempt to manually script AGServiceManager or
NodeManagementService etc to run on this (the lead) machine - always let
the VenueClient start personal node services as needed for you (it will
actually start AGServiceManager and NodeManager(?) for you).
c. On all other machines, set AGServiceManager to run as default through
whatever scripting system you want to use. Only AGServiceManager is
needed. Its best to create long proxies on these machines for this
reason, as these services will rarely be used interactively later. No
command line options need be specified.
d. With AGServiceManager running on the slave machines, start up
VenueClient --personalNode on your lead machine.
e. Chose the
VenueClient > preferences > Manage my node
option, and the node managment app should pop up. by default the
localhost will be in there with little else. you should be able to choose
Service manager > Add..
from the menu and add the other two service managers by IP address.
(Only AGServiceManager.py needs to be running on these to make this
choice work).
f. If you get this far successfully, and have three service managers
listed, you are home and dry. Add/remove and edit your services as
required across all three machines. when you are happy, choose
File >Store Configuration
from the Node Management app.
You can choose the default configuration, and check use as default, and
all your changes should stay put, for all users of the machine.
In future, always boot all three machines. make sure AGServiceManager
runs as standard on the two slave units. On the main lead unit, simply
run VenueClient with personalNode services enabled - on windows just
click the desktop shortcut, as --personalNode is already included in that.
The personal Node charaterisation is a bit of a misnomer in this for
sure, but no one sees it. And this way the config stays put over
reboots and correctly starts required services on each unit automatically.
Other routes will see your config and services choices disappearing
after reboots (if you get that far) and its a right work up.
Hope this helps get you going sooner rather than later, or at least
shakes something concrete loose to debug.
bye
lewis
--
Lewis Grantham LLB MSc
Project Leader
Media Resources (Multimedia Unit)
Windeyer Building
Cleveland St
London
W1T 4JF
Tel:0044(0)2076799258
Fax:0044(0)2075800995
More information about the ag-tech
mailing list