- Oct 9, 1999
- 21,019
- 156
- 106
With RC5 winding down I want to start thinking about what I might switch over to. Because of how I have to run things, not every project makes sense.
My herd has no net access. I have a pproxy internally and the herd connects to it to fetch/flush, but the pproxy can't get its own blocks or flush them. I have a personal proxy at home which I have to use to get a week's worth of blocks, email them in, and manually set up the file for the pproxy running the herd. Flushing the pproxy is the same way - email the finished blocks home, fiddle with the files for the home pproxy, then flush them.
Right now I have about 40 machines (90% Pentium 2 at 266mhz or 300mhz, and 10% Pentium 4's). But in a few months, all the machines will probably become Pentium 4's at about 1.8ghz. Once that happens, I'm probably going to be just on the brink of not being able to get enough RC5 blocks in and out via email. I would always have the option of just not using all available machines, but that's not very appealing.
Using CD-Rs is possible so I'll edit my original post to clarify that also. My only concern there is being able to fetch and flush work in a reasonable amount of time. Right now I can fetch and flush a week's worth of work in an hour or two from home (56k dialup), which is fine. Something in the range of 2-4MB a couple times a week is about as much as I could manage. Even if I could use a CD-R to move the work around, I can't take three hours a day to download/upload.
So I'm wondering, what other DC projects do this:
If I knew which DC projects worked that way, I could start researching more details now while I have time to plan.
Thanks in advance for your help.
My herd has no net access. I have a pproxy internally and the herd connects to it to fetch/flush, but the pproxy can't get its own blocks or flush them. I have a personal proxy at home which I have to use to get a week's worth of blocks, email them in, and manually set up the file for the pproxy running the herd. Flushing the pproxy is the same way - email the finished blocks home, fiddle with the files for the home pproxy, then flush them.
Right now I have about 40 machines (90% Pentium 2 at 266mhz or 300mhz, and 10% Pentium 4's). But in a few months, all the machines will probably become Pentium 4's at about 1.8ghz. Once that happens, I'm probably going to be just on the brink of not being able to get enough RC5 blocks in and out via email. I would always have the option of just not using all available machines, but that's not very appealing.
Using CD-Rs is possible so I'll edit my original post to clarify that also. My only concern there is being able to fetch and flush work in a reasonable amount of time. Right now I can fetch and flush a week's worth of work in an hour or two from home (56k dialup), which is fine. Something in the range of 2-4MB a couple times a week is about as much as I could manage. Even if I could use a CD-R to move the work around, I can't take three hours a day to download/upload.
So I'm wondering, what other DC projects do this:
- works without net access
- one pproxy to serve the herd
- easy to get a week's worth of work back and forth via email and the files can't be huge. For RC5, I can sometimes get a week's worth of blocks (60,000+) in a file that's under 1MB. I can't live with 5MB files that I have to send back and forth every day. [edit: CD-Rs are possible but I have to use a 56k dialup to download/upload, so I can't afford three hours a day to keep things going]
- client runs as a service under NT (nothing in the systray) (pproxy doesn't have to be a service)
- client never requires attention
- has a tool like the cool rc5desstats program that scans the pproxy log files and gives me stats by machine (I guess I could live without this)
If I knew which DC projects worked that way, I could start researching more details now while I have time to plan.
Thanks in advance for your help.