From: xie xingguo <xie.xingguo@zte.com.cn>
Date: Fri, 23 Feb 2018 03:39:13 +0000 (+0800)
Subject: osd/PG: decay scrub_chunk_max too if scrub is preempted
X-Git-Tag: v12.2.6~98^2~13
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=cdbf3242dc8921f534ec93e4ee5517460f7ba36f;p=ceph.git

osd/PG: decay scrub_chunk_max too if scrub is preempted

In normal case we'll at least scrub as many objects as
osd_scrub_chunk_max specified at a time, so the current
backoff mechanism should have very limit effect.
Decay both osd_scrub_chunk_min and osd_scrub_chunk_max
should instead be a better resolution.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
(cherry picked from commit a9260524676ac28742e5a945de93b87ae985017e)
---

diff --git a/src/osd/PG.cc b/src/osd/PG.cc
index 0641e59913e8..f4144cb7fe5b 100644
--- a/src/osd/PG.cc
+++ b/src/osd/PG.cc
@@ -4662,7 +4662,8 @@ void PG::chunky_scrub(ThreadPool::TPHandle &handle)
 	   */
 	  int min = std::max<int64_t>(3, cct->_conf->osd_scrub_chunk_min /
 				      scrubber.preempt_divisor);
-	  int max = std::max<int64_t>(min, cct->_conf->osd_scrub_chunk_max);
+	  int max = std::max<int64_t>(min, cct->_conf->osd_scrub_chunk_max /
+                                      scrubber.preempt_divisor);
           hobject_t start = scrubber.start;
 	  hobject_t candidate_end;
 	  vector<hobject_t> objects;