mgr/balancer: fix fudge

next_* simply record some intermediate results when balancer tries
to find something to optimize, whereas best_* are used for the
final outputs.

Fix by comparing origin_ow with best_ow when determining if we
should use a valid fudge value.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
(cherry picked from commit 8f46a351f1fa6f921b373aed4038efc0898a8a44)

diff --git a/src/pybind/mgr/balancer/module.py b/src/pybind/mgr/balancer/module.py
index 80692929c2b071d5608b738dadb51e8d0a3d1e0e..7868cd2fbc174e3f48265944961c768d27962cb2 100644 (file)
--- a/src/pybind/mgr/balancer/module.py
+++ b/src/pybind/mgr/balancer/module.py
@@ -1125,7 +1125,7 @@ class Module(MgrModule):
 
         # allow a small regression if we are phasing out osd weights
         fudge = 0
-        if next_ow != orig_osd_weight:
+        if best_ow != orig_osd_weight:
             fudge = .001
 
         if best_pe.score < pe.score + fudge: