Clear JRE cache remotely
Hi,
I have a mass deployment problem : I need to clear the JRE cache on several computers remotely (versions 1.3.1 and 1.5.0).
Does someone knows if I can invoke the clear action remotely without having to take the control of the computer desktop ?
Maybe by using an API with a distributed executable for example ?
Thanks.
Rgs.
[364 byte] By [
armoreta] at [2007-11-27 10:54:19]

# 4
The real problem is that, for an unknown reason, some applet upgrades don't work correctly unless clearing the JRE cache (that makes reloading the whole code that is spread into multiple jars, etc...).
As I don't have the possibility to investigate the real problem origin and I'm not sure at all to find out the solution, as the problem happens on more that 500 workstations (stations without any users behind), all I can do is to investigate how applicate the side around solution that I know is working for sure.
Please, -please-, don't ask why it doesn't work : just answer to the question ! (but related suggestions still welcomed of course)
Rgds.
# 5
What is happening is browsers cache applets client-side, but if the server-side applet is changed, the browser will not download a new one. If the applet is removed from the browsers cache, it will be forced to download a new applet. However, it seems like the article below is helpful.
http://java.sun.com/products/plugin/1.3/docs/appletcaching.html
Long story short, in the web page HTML, use an EMBED or OBJECT tag and then a
<PARAM NAME="cache_option" VALUE="no">
will make sure the applet is downloaded every time the page is loaded. Also, it seems like using the line
<PARAM NAME="cache_option" VALUE="Plugin">
will cause the applet on the client to be automatically updated when the applet on the server is updated.
# 6
Interesting suggestion, I will check if this can be applied in my case. Obviously the objective is not to reload the applet every time but only if one of its files is updated (can be any of its jars). Without any explicit directive, the latest cache option you describe is not supposed to be the default behaviour ?