Posts by Snider

1) Message boards : Number crunching : Scheduler request failed: HTTP internal server error (Message 67059) Posted 1 Aug 2010 by Snider Post: same old firewall & virus. even disabled both firewall & virus without any change. Repeating my post below, exiting & restarting BOINC resolved issue for unkown reason.
2) Message boards : Number crunching : Scheduler request failed: HTTP internal server error (Message 67058) Posted 1 Aug 2010 by Snider Post: Exiting BOINC & restarting solved it -- rather unsettling!
3) Message boards : Number crunching : Scheduler request failed: HTTP internal server error (Message 67042) Posted 30 Jul 2010 by Snider Post: 7/29/2010 11:07:17 PM rosetta@home Sending scheduler request: To fetch work. 7/29/2010 11:07:17 PM rosetta@home Requesting new tasks 7/29/2010 11:08:27 PM rosetta@home Scheduler request failed: HTTP internal server error Consistantly receiving above error. Other machines on same network are working fine. Resetting project has no affect. Anyone else have this server side problem?
4) Message boards : Number crunching : still unable to merge (Message 50253) Posted 2 Jan 2008 by Snider Post: It has always been that way. There is less chance of problems this way than trying to save the old number. Host number, I can understand. However, the OLDEST creation date should be maintained so users knows when that host originally joined project.
5) Message boards : Number crunching : still unable to merge (Message 50174) Posted 29 Dec 2007 by Snider Post: It appears that the oldest created host is LOST when merging. It would have been nice after the merge to have the hosts show the ORIGINAL create/join date rather than the most recent.
6) Message boards : Number crunching : Diskless crunching methods (Message 49742) Posted 17 Dec 2007 by Snider Post: I have a no hard drive single-core older unit w/ 512MB RAM that boots from CD-R. WinXP boot using boincPE. boincPE (based on BartPE) is completely free, though not for a newbie to setup. Files are stored on another networked computer's HD (which also supports boinc). It took me a number of CD-R coasters before I got one fully working. ENGLISH details on BOINCpe: http://blog.schreiter.info/boincpe-livecd-for-boinc/ As I recall ... different computer names need different copies of the CD. The files stored on the network drive are based on MAC. Web site suggests using BAM to make initial setup & configuration easier. The 2nd boincpe machine I setup died before I got a stable setup that was able to return completed WUs. If I setup another (or have to redo the 1st), I'd try to get ALL of BOINC on network HD vice on the CD. That would be MUCH easier to upgrade BOINC. It wasn't until seting up boincPE that someone pointed out the feature in BOINC Manager / Advanced / Select_Computer. It allows you to connect to boinc running on another machine to see what is going on and change configurations. One gotta: if boincPE's connection to network drive is lost, then boincPE does not allow shutdown/reboot because it cannot write files to lost network drive. I had to add another batch file to "shutdown -r" without writing current data to hard drive.
7) Message boards : Number crunching : Settings that might help the R@H servers (Message 45977) Posted 11 Sep 2007 by Snider Post: Given an always-on, dedicated Rosetta cruncher, is there any reason not to walk the the Connect to Network time out to 3 or 4 days? Personally/morally, I think the larger your cache (extra WUs download to process in the future), the larger your RESPONSIBILITY to actually complete and upload before the deadline. I have no bias for or against any project THAT I'M ATTACHED TO. But I am concerned that one of my machines might hiccup or crash, especially when I am gone for days at a time. Or when the machine is in another city. So for me, I'm most comfortable only downloading WUs that I'm going to work on NOW, finish them, then get more WUs. Even if that means other projects get more CPU cycles. I always strongly recommend to folks to attach to at least 2 projects. Can make the share resources extremely unequal (100,1). Such that the second project will mostly not do anything unless you run out of work or the primary project hiccups (like R@H did recently) or runs out of work (like Predictor@Home has done).
8) Message boards : Number crunching : Using University Computers (Message 45974) Posted 11 Sep 2007 by Snider Post: Perhaps this will help. The central computing department on campus operates many thousands of machines and recently deployed R@H. I've found it's easier to convice folks to copy what others are doing rather than starting from scratch. http://www.washington.edu/nebula/boinc.html Does UW allow, prohibit, encourage, discourage UW folks who are running BOINC with R@H to also run other BOINC projects? Rival schools to UW may be enticed by the challenge to process more WUs/credit than UW does?!!