Hi - I personally see all of the development happening interdependently. I don't think it would be a good idea to implement phase 1 without doing the groundwork for phase 2 at the same time. I feel the same way about making sure a tool like bittorrent can plug in to the transport module. I think it needs to be tested along with the http implementation. The thought I had around two phases was not trying to deliver everything at once more so than only working on one piece at a time.
thx, ginnie On 05/22/09 12:59, Sundar Yamunachari wrote: > Dave Miner wrote: >> Sundar Yamunachari wrote: >>> Dave Miner wrote: >>>> >>>>> We discussed the Questions to be answered on the Problem >>>>> statement >>>>> page. They have been updated. We determined that we will implement >>>>> the >>>>> mechanism as Phase 1 of the project. >>>>> Phase 2 of the project will consist of implementing an >>>>> observability tool. We would also like to investigate bittorrent as >>>>> another module. This would provide us a proof of concept that >>>>> the mechanism we've written is modular. >>>>> >>>> >>>> I'm pretty negative on the idea of implementing observability >>>> later. Providing good observability is typically something >>>> reflected as a core aspect of a design, not something to be glued >>>> on later. >>> We are talking about the server side monitoring tool that which can >>> observe more than the transport. Our design will include logs, >>> proper error message, status on the client and server side. How we >>> use those discrete information to provide a unified message to the >>> user in not clear yet. >>> >> >> I don't see anything there that indicates why it would not be a >> candidate for work during an initial phase. > We talked about sending status back to server like SNMP traps, > transferring the logs (in the case of failure) etc. We have to decide > what facilities are available to interpret the information coming from > the client. Some of the observability discussion is captured in > http://www.opensolaris.org/os/project/caiman/auto_install/WebServerProblemStatement/. > > We also discussed some stages like failures after solaris is booted > are more observable than others like failures during pxe boot. Each > stage of the sparc and X86 booting is captured in > http://www.opensolaris.org/os/project/caiman/auto_install/WebserverUseCases/ > > > - Sundar >> >> Dave >> > -- Ginnie
