From: Iustin Pop <iustin@google.com>
Date: Sun, 15 Feb 2009 13:53:38 +0000 (+0100)
Subject: Documentation updates
X-Git-Tag: htools-v0.0.4^0
X-Git-Url: https://code.grnet.gr/git/ganeti-local/commitdiff_plain/d53264c02f36aec33ed05ffd27bfd88c80254546

Documentation updates
---

diff --git a/README b/README
index b4e5711..03f3209 100644
--- a/README
+++ b/README
@@ -112,11 +112,15 @@ Command line usage
 
 Synopsis::
 
-    hn1 [-n NODES_FILE] [-i INSTANCES_FILE] [-d START_DEPTH] \
+    hn1 { [-n NODES_FILE] [-i INSTANCES_FILE] | [-m CLUSTER] } \
+        [-d START_DEPTH] \
         [-r MAX_REMOVALS] [-l MIN_DELTA] [-L MAX_DELTA] \
         [-p] [-C]
 
-The -n and -i options change the names of the input files. The -d option
+The -n and -i options change the names of the input files.
+Alternatively, the -m option specifies collection of data via RAPI.
+
+The -d option
 changes the start depth, as a higher depth can give (with a longer computation
 time) a solution with better delta. The -r option restricts at each depth the
 number of solutions considered - with r=1000 for example even depth=10 finishes
@@ -211,22 +215,30 @@ Command line usage
 
 Synopsis::
 
-    hbal [-n NODES_FILE] [-i INSTANCES_FILE] \
+    hbal { [-n NODES_FILE] [-i INSTANCES_FILE] | [-m CLUSTER] } \
          [-r MAX_ROUNDS] \
-         [-p] [-C]
+         [-p] [-C] [-o]
+
+The -n and -i options change the names of the input files.
+Alternatively, the -m option specifies collection of data via RAPI.
 
-The -n and -i options change the names of the input files. The -r option
-restricts the maximum number of rounds (and is more of safety measure).
+The -r option restricts the maximum number of rounds (and is more of
+safety measure).
 
 The -p option will show the cluster state after the solution is implemented,
 while the -C option will show the needed gnt-instance commands to implement
-it.
+it. The -o option specifies that instead the default, quite verbose
+output, a single line of output should be shown, in the format::
+
+  initial_score number_of_moves final_score improvement
+
 
 Integration with Ganeti
 -----------------------
 
-The programs needs only the output of the node list and instance list. That is,
-they need the following two commands to be run::
+The programs can either get their input from text files, or directly
+from a cluster via RAPI. For text files, the following two commands
+should be run::
 
     gnt-node list -oname,mtotal,mfree,dtotal,dfree \
       --separator '|' --no-headers > nodes
@@ -235,6 +247,9 @@ they need the following two commands to be run::
 
 These two files should be saved under the names of 'nodes' and 'instances'.
 
+For RAPI, the "-m" argument to both hn1 and hbal should specify the
+cluster or master node name.
+
 When run, the programs will show some informational messages and output the
 chosen solution, in the form of a list of instance name and chosen
 primary/secondary nodes. The user then needs to run the necessary commands to
diff --git a/src/Cluster.hs b/src/Cluster.hs
index 1cf5153..e8430d6 100644
--- a/src/Cluster.hs
+++ b/src/Cluster.hs
@@ -431,7 +431,6 @@ checkMove nodes_idx ini_tbl victims =
              if null vtail then best_tbl
              else checkMove nodes_idx best_tbl vtail
 
-
 {- | Auxiliary function for solution computation.
 
 We write this in an explicit recursive fashion in order to control