What's with all the boinc projects with HW issues?

An inspired rant from seeing things with Dirt, Milkyway, SETI, etc....

I just don't get all these projects with server issues due to hardware.

most boinc servers do not need to be some uber, complicated machine. Heck, many can be run on a VM under Windows or OS X. Granted some more complicated projects use more than one server.

A 2.6ghz Quad core (or a faster 6-8 core), 4-8GB of ram, cheap gpu for display.. monitor optional (as you can telnet/ssh into linux and pretty much do anything needed), 0.5-1TB HD should be more than enough for a boinc server. All that can be purchased for probably <750$

If people are worried about uptime, keep a spare, identical mobo on hand, spare identical cpu, a CHEAP identical gpu on hand, and a spare HDD on hand... ram... you can order and get in 24 hrs or in a pinch run to best buy (bleh).. oh and a backup solution (usually a usb HD of equal or greater size than the internal storage). Basically just keep a duplicate of the server. If the server acts up due to HW, just slap the HD(s) into the backup box. Since everything is identical, the server should just boot and continue normally. If the HD(s) fail, restore the backup to the backup box and fire stuff back up. Personally, I'd only run a boinc server on a striped raid config so if one HD fails, it can be replaced w/o data loss.

It doesn't require a person to be rich to do any of this. Plus hardware running 24/7 is better for it anyway (as long as it's properly cooled). I also don't get these projects with overheating servers... Seriously. Are they running them in a 75+F room with the servers and fans clogged with dust? I just don't see a boinc server's CPUs being hammered like the boinc clients hammer client cpu/gpus. As long as you don't OC the CPU, a computer should be able to cool itself properly, under load, even in a room that's 80-85F.

I guess if collage students are running projects.. well then 750$ for a machine and 750 more for backup HW would be a lot. I just know how I'd do it. That way if any HW issues crop up, downtime is only a few hours. And when HW issues should keep plaguing a server, to me, that means either the server is being run in a poor environment, not being cared for properly, and/or it's using aging and/or used donated HW.

Now software issues with boinc I understand... lots to go wrong when you compile the server yourself... code the WUs yourself, etc... So I get issues in that department... and there can be a LOT of them which can cause server issues, downtime etc... but that's all SW and not HW

I just do not get these projects plagued with HW issues...

Hmmm...I don't know enough about running a project, but I agree to a point - notice how Collatz (Slicker's project) is hardly ever down?:cool:

Lol yep.. And Yafu is a vm and a more stable than many other projects