Data/Code file distribution/patch possibility with Ice

[spsoni]

Hi,

I frequently need to transfer files (data or code) to the server from client. Is IcePatch a good idea?

My understanding with IcePatch is that, we need to create the patch first then start the icepatch server then run icepatch client on destination node and get the patch. If there are updates in the distribution, we need to shutdown the server, recreate the patch (only updates for updated files) and start the icepatch server again and run the client again.

In my scenario, I will need to frequently update/create files. So is it possible to automate all above process every time I execute my client? Can we programatically create patch, start/stop patch server and trigger destination node servers to fetch the updates from patch server?

I am open to alternative framework/solution to this problem.

I am using FreeBSD 7.0 and Ice 3.2.1 (Yes, I am still waiting for next patch for Ice 3.3.0 on FreeBSD, I will soon upgrade myself)

Many thanks.

[benoit]

Hi,

You can programatically start and stop the IcePatch2 server if it's deployed with IceGrid but you can't easily re-create the checksums for a new file set programatically: you have to run icepatch2calc while the IcePatch2 server is down. Providing the possibility to update the IcePatch2 server without a restart is on our long TODO list... if you have a commercial interest for this feature we could give this a higher priority, please contact us at info@zeroc.com if that's the case.

Also note that if you use IceGrid distributions with IcePatch2, you can easily trigger an update of the IceGrid nodes with the application or server patch IceGrid admin commands.

Finally, if your file set is often updated it might be best to implement your own custom file transfer mechanism. For information on how to implement efficient file transfer, I recommend taking a look at Matthew's newsletter article here.

Cheers,
Benoit.