Just a minor update to announce that javadocs for the remote-ws and remote-gui components are now available online. The web service documentation is so far the most polished one, and should serve as a good introduction to some of the client server interaction.
Also, I have started a wiki page with some enhance proposals for the next major milestone. Right now it is just a simple "scratch" page with loose ideas, but hopefully it will evolve into a useful resource for planning the future development and collaboration. Expect REPs (Re-Mote Enhancement Proposals) to land in the future!
Re-Boot Testbed
A lot of things have happened in the last week. In preparation for the upcoming release, all the repositories have been tagged as version "1.0pre0", however no archives have been uploaded to the project download area. Some minor cleanups have also managed to sneak in. Other project updates include closing of some more issues related to Milestone1.0.
In other news, installation of the DIKU Testbed server is almost complete and a small test setup is already up and running in my office with PXE booting mote hosts. I probably won't have time to make the first test deployment before Thursday or Friday, however, with the hardest parts of the reinstall out of the way the rest should be over quickly. I hope to get time to write down a walk-through or howto covering the basics of what I did either here in the blog or at the project wiki.
Related to the reinstall, I made an effort to try and strip down the webapps/axis directory from axis-1.0. The idea is of course to be able to provide an easily deployable web application archive (.war file) complete with web service descriptors and associated jar files, as well as a simple administrator interface and dependency validation similar to that offered by the example axis web application. It is still a work in progress in terms of integration, but I feel it is a good start.
In other news, installation of the DIKU Testbed server is almost complete and a small test setup is already up and running in my office with PXE booting mote hosts. I probably won't have time to make the first test deployment before Thursday or Friday, however, with the hardest parts of the reinstall out of the way the rest should be over quickly. I hope to get time to write down a walk-through or howto covering the basics of what I did either here in the blog or at the project wiki.
Related to the reinstall, I made an effort to try and strip down the webapps/axis directory from axis-1.0. The idea is of course to be able to provide an easily deployable web application archive (.war file) complete with web service descriptors and associated jar files, as well as a simple administrator interface and dependency validation similar to that offered by the example axis web application. It is still a work in progress in terms of integration, but I feel it is a good start.
Reinstallation Count-down
The last few days have been a bit slow with no real progress on the testbed. I've spoken with Janus, the engineer at DIKU, about how we should proceed and he has provided the missing hardware pieces for what will become testbed.diku.dk. With that out of the way, I hope to begin setting up a small test environment in my office by the end of the week. It will be good to have the system running again; not being able to properly test is not very inspiring.
While wasting my time on the great internet the other night I again stumbled upon Google Web Toolkit. It looks good, quite easy to use, and has a nice framework for testing with both command line support and plugins for both NetBeans and Eclipse. I think it will fit all the needs of the admin interface and help to keep as much as possible of the high-level server code in one language. Though I still need to read more about how the server-side part of the toolkit works, the dream of a future web service module with easy installation and a nice admin interface lives on. It will also be fun to see if the database part can be merged into the web service module so that site specific costumization "plugins" defining new web services will be possible.
[ BTW, I played with Yahoo pipes yesterday and created the "Recent Changes" RSS feed displayed in the right column. It aggregates commit activity from the four DIKU git repositories on repo.or.cz. ]
While wasting my time on the great internet the other night I again stumbled upon Google Web Toolkit. It looks good, quite easy to use, and has a nice framework for testing with both command line support and plugins for both NetBeans and Eclipse. I think it will fit all the needs of the admin interface and help to keep as much as possible of the high-level server code in one language. Though I still need to read more about how the server-side part of the toolkit works, the dream of a future web service module with easy installation and a nice admin interface lives on. It will also be fun to see if the database part can be merged into the web service module so that site specific costumization "plugins" defining new web services will be possible.
[ BTW, I played with Yahoo pipes yesterday and created the "Recent Changes" RSS feed displayed in the right column. It aggregates commit activity from the four DIKU git repositories on repo.or.cz. ]
Updating and Reorganizing the Wiki
The last couple of days, I've been working on updating the wiki pages. A lot of the changes went into reorganizing the manual for administrators by combining the individual pages and improving and cleaning it all up. It is not complete yet, but should give a better idea about dependencies and other requirements. I also made the source code page more simple and less "pro-git" as well as listing all the repositories more uniformly. Finally, I moved the user manual to Google Docs.
Generally, I like doing documentation updates and find that documentation work is a good way to get (back) into a project: it is usually always a place that needs some extra attention and it quickly fills the head with new ideas. Of course, it is also worth noting that it is considerably harder to create real havoc when making changes to documentation than code.
Working on the administrator manual got me thinking whether we should consider (re-)packing the axis web service archive so that administrators can simply unzip a prepared and working remote-server and be up and running with less work. It could have a nice test page, web service deployment configuration, and maybe even an database installation/upgrade interface. This will be even more relevant when the admin interface will become reality, and is definitely something I would like to look more into. There might of course also be some problems with such a "drop-in" solution, since it can make it harder for administrators that make a lot of site specific configurations of the testbed application.
Generally, I like doing documentation updates and find that documentation work is a good way to get (back) into a project: it is usually always a place that needs some extra attention and it quickly fills the head with new ideas. Of course, it is also worth noting that it is considerably harder to create real havoc when making changes to documentation than code.
Working on the administrator manual got me thinking whether we should consider (re-)packing the axis web service archive so that administrators can simply unzip a prepared and working remote-server and be up and running with less work. It could have a nice test page, web service deployment configuration, and maybe even an database installation/upgrade interface. This will be even more relevant when the admin interface will become reality, and is definitely something I would like to look more into. There might of course also be some problems with such a "drop-in" solution, since it can make it harder for administrators that make a lot of site specific configurations of the testbed application.
The DIKU Testbed Server Went Offline
Yesterday, the old DIKU Testbed server was taking offline. It has been living in a server room near the old Distlab area. The room contains machines from various research groups and student projects and has traditionally been managed in a very ad-hoc manner. I had a hard time identifying which machine it was, so powering it off (something that had to be done eventually) sounded like a good option. I wouldn't say it worked too well, at least I was not completely sure that I had found the right box, but Esben Zeuthen came to my rescue and helped me locate it on the top shelf.
I will definitely need to get a ladder next Monday, when I have to take the machine down for reparations. The harddisk has not been working too well for the past few months. With a new disk, I will also have a much needed opportunity to upgrade the system to the latest Debian or Ubuntu. Besides a new hard disk, a new network interface since the architecture for the new testbed will put all the mote hosts on their own separate network. This understandably makes the system administrators at DIKU a lot less worried and will allow us to have the mote hosts boot over the network, making future upgrading much easier. I hope to make progress on the new infrastructure next week.
Finally, mail aliases and a DIKU subdomain name seem to also have been sorted out. Questions regarding DIKU Testbed can be mailed to testbed@diku.dk and the new box will be available at testbed.diku.dk instead of amigos30.diku.dk.
I will definitely need to get a ladder next Monday, when I have to take the machine down for reparations. The harddisk has not been working too well for the past few months. With a new disk, I will also have a much needed opportunity to upgrade the system to the latest Debian or Ubuntu. Besides a new hard disk, a new network interface since the architecture for the new testbed will put all the mote hosts on their own separate network. This understandably makes the system administrators at DIKU a lot less worried and will allow us to have the mote hosts boot over the network, making future upgrading much easier. I hope to make progress on the new infrastructure next week.
Finally, mail aliases and a DIKU subdomain name seem to also have been sorted out. Questions regarding DIKU Testbed can be mailed to testbed@diku.dk and the new box will be available at testbed.diku.dk instead of amigos30.diku.dk.
Starting work on DIKU Testbed v2.0
After finally getting installed in the new offices (actually it is where DIKU initially started) work on setting up and relaunching the DIKU Testbed has begun. If everything goes well, testing can begin next week with three mote hosts and around 12 motes.
The purpose of these preliminary tests is to see if the communication of the motes in the testbed affects the wireless network at DIKU. In the previous deployed testbed, the motes often received junk packages, but not at a rate where it severely damaged their connectivity. It will be interesting to see if I can get some measures of the numbers for dropped packages etc. on the motes.
Subscribe to:
Posts (Atom)