Uninstalling does not loose the work... it leaves the data directory and upon reinstall the project should be showing as already attached, so add that to the riddle. It's not when you go into the desktop screensaver applet the reference to the BOINC is being slipped in? I've got it on Empty SS, 5 minutes.

ZA, to some the Kaspersky of firewalls. Set all 5 BOINC apps (boinc.exe, boincmgr.exe boinccmd.exe, boincscr.exe and boinc.scr) to freepass localhost and port 31416 plus ports 80 + 443 for grid server connection. In the AVG exempt the BOINC data_dir from scanning. Speed up things too.

Hi.

I've got a question about 6.10.58 for Ubuntu/Linux, has it got a better work fetch scheduler than 6.10.17 because this thing seems screwy.

It downloads nothing from one project then next time it gets 30 + tasks and nothing for the other projects, then does the same thing to another project not a few at a time.

I would like to see work for all projects on board and yes my all projects have an equal share.

I will upgrade if it's any better.

The jury is out on that on a permanent basis, as when there's no hope to give you any sound comment without knowing all of your settings (Start Log, first 30 lines, and if you use local prefs, the global_prefs_override.xml content) and the weights/resource shares of the projects as seen in the project tab of the BM. My Linux quad with 6.10.58 from GetDeb goes out on regular basis to backfill the cache. Occasionally when a task runs much faster than the current TTC estimates, it fetches some more in one call such as just now when the C4CW as grossly overestimated at 5:53 hours, but really needing only 1:25. On a 2 day cache that already causes considerable cache TTC expanding and shrinking with each completed task. You'd also have to look at debts. Why erratic behavior to the casual observer?: Either one digs into the documentation, help FAQ's to understand or one lets BOINC alone and does not change the settings every other moment. BOINC needs weeks to figure out what is done, how long work takes e.g. so it can apply that knowledge to project CPU time allocation.

And, just reading the alpha mail list, it's still up to it's old "try starting many" trick with 6.10.58 when there is a panic state (too much cached). Caches of 1-3 days for those with good connection is more than sufficient. Anyone that caches more and constantly tweaks is doing it to her/himself... hit the books! The client defaults are there for a reason and are mostly fine.