[Swift-commit] r4163 - SwiftApps/SwiftR

[noreply at svn.ci.uchicago.edu]

Author: tga
Date: 2011-03-02 16:34:02 -0600 (Wed, 02 Mar 2011)
New Revision: 4163

Modified:
   SwiftApps/SwiftR/IMMEDIATE-TODO
Log:
Updates to swiftR todo list.


Modified: SwiftApps/SwiftR/IMMEDIATE-TODO
===================================================================
--- SwiftApps/SwiftR/IMMEDIATE-TODO	2011-03-02 22:29:29 UTC (rev 4162)
+++ SwiftApps/SwiftR/IMMEDIATE-TODO	2011-03-02 22:34:02 UTC (rev 4163)
@@ -14,12 +14,9 @@
     - smaller focused micro-tests looking at param size
     - Sarah Kenny FMRI
     - parallel OpenMx Tests
-    - Candidate systems:
-        - Eureka/cobalt - need to finish up
-        - Beagle - might be working
-        - Sicortex (NO)
-        - BG/P (NO)
 
+HIGH:
+-- get working on beagle.
 
 HIGH:
 - tests
@@ -35,14 +32,15 @@
  -- swift times?
  -- micro studies on provider staging etc.
 
-HIGH:
 
-- test on Ranger
-
 HIGH:
 See email about pbs breaking on UVa cluster
 
 MID:
+
+- test on Ranger
+
+MID:
 - saner approach to channels: channel per request to avoid the issue
     of what happens if a "done" is never read
 
@@ -60,9 +58,6 @@
  -- siraf
  -- godzilla (if updated)
 
-MID:
-- user testing in general (ssh, pbs, sge) 
-* SGE testing: ranger, siraf (low priority)
 
 MID:
 * Cleanup ssh worker processes: add a watchdog that detects when worker.pl has 
@@ -74,13 +69,13 @@
 
 MID:
 * Fast branch: working? faster?
+    (* NOTE: wait until updates: fast branch afflicted by deadlocks *)
 
 MID:
-Investigate use of swift broadcase:
-    swift bcast if exportAll files/data and/or xmit this via prov staging w/ caching of duplicates
+Implement omxXXX parallel calls, update openmx code
 
 MID:
-Implement omxXXX parallel calls, update openmx code
+- More robust mechanism for swiftExport - use swift to stage?
 
 LOW: (unless needed by immediate OpenMx app or test)
 - complete sf compat functions (sapply, lapply -> for openMx, based on usage)
@@ -94,7 +89,6 @@
       - SwiftR removes result pipe
       - Swift script writes a done message to result directory
         and then exits loop
- -- workspace is run as a job under batch scheduler -> new feature
 
 LOW: 
 -- perf issue wrt arg list copies (very low prio)
@@ -115,7 +109,11 @@
     - each request provides a reply channel
     - swift server forks off a worker thread for
         each request
+LOW:
+Investigate use of swift broadcase:
+    swift bcast if exportAll files/data and/or xmit this via prov staging w/ caching of duplicates
 
+
 =========================================================================
 Completed:
 =========================================================================
@@ -139,3 +137,12 @@
     He (Mihael) will also look at a better fix to the coaster timeout problem, but for now, you should integrate the timeout change from my trunk/cog/modules/provider-coaster/src/* into your test trunk/
     Otherwise, you'll find that your coaster workers quit after a few minutes of inactivity and then start-swift needs to be killed, workers cleanup up, and start-swift restarted.
 
+MID:
+- user testing in general (ssh, pbs, sge) 
+* SGE testing: ranger, siraf (low priority)
+
+LOW:
+ - workspace is run as a job under batch scheduler -> new feature
+ (* This can be achieved by running R script through batch scheduler, and using
+ ssh server.  A convenience function getNodeList is included to make it easier
+ to get the host list from the environment *)