might be me, I have been clicking wild on purpose. Folker On Apr 24, 2008, at 3:14 PM, Robert Olson wrote:

I think we may have to put any analysis in the mgrast into short- term batch mode. with what I'm guessing is just a couple of us using the system I'm seeing this:

14723 olson 25 0 8848m 2.8g 2.8g R 100 18.1 13:58.04 postgres 15164 olson 25 0 8828m 1.2g 1.2g R 100 7.9 1:18.13 postgres 14509 olson 25 0 8856m 3.0g 2.9g R 100 19.0 24:34.44 postgres 14510 olson 18 0 8841m 3.0g 3.0g R 89 19.0 24:36.10 postgres 15163 olson 25 0 8843m 1.3g 1.3g R 86 8.2 1:46.44 postgres 14921 olson 25 0 8933m 6.2g 6.1g R 67 39.9 7:51.13 postgres 15058 olson 25 0 8833m 829m 813m R 65 5.2 4:07.23 postgres 14724 olson 25 0 8856m 2.8g 2.8g R 50 18.1 13:28.68 postgres 15060 olson 21 0 8833m 834m 818m R 50 5.2 4:14.50 postgres 15080 olson 25 0 8835m 1.4g 1.3g R 46 8.6 4:03.44 postgres 14943 olson 25 0 8924m 5.8g 5.8g R 35 37.3 7:00.00 postgres

Given the cost of these queries I'd also like to see the ajax stuff not kick off automatically when you go to an analysis page; rather, the user has to explicitly choose parameters (even the defaults) and request the analysis.

I'd also like to reiterate here in this context that we shouldn't run multiple analyses in parallel via ajax from one of these pages.

--bob