]> git.apps.os.sepia.ceph.com Git - ceph.git/commitdiff
common/options/global.yaml.in: increase default value of bluestore_cache_trim_max_ski... 40920/head
authorNeha Ojha <nojha@redhat.com>
Thu, 15 Apr 2021 16:44:27 +0000 (16:44 +0000)
committerNeha Ojha <nojha@redhat.com>
Mon, 19 Apr 2021 17:39:46 +0000 (17:39 +0000)
This option controls the rate of trimming of onodes and the earlier default of
64 has been seen to be too low for large clusters, leading to buildup of
onodes resulting in memory growth.

Increase the default value to 1000, since there are no known downsides to it.

Fixes: https://tracker.ceph.com/issues/50217
Signed-off-by: Neha Ojha <nojha@redhat.com>
(cherry picked from commit 26d7336d6b65f987298ede5d2c5c435191f1405c)

 Conflicts:
src/common/options/global.yaml.in - file does not exist in nautilus

src/common/options.cc

index 768d6505d9d876a575e35e79e4722ea7932c3297..5524da48d1d2d6118eae99b25a8e423bf4bea8d6 100644 (file)
@@ -4744,7 +4744,7 @@ std::vector<Option> get_global_options() {
     .set_description("How frequently we trim the bluestore cache"),
 
     Option("bluestore_cache_trim_max_skip_pinned", Option::TYPE_UINT, Option::LEVEL_DEV)
-    .set_default(64)
+    .set_default(1000)
     .set_description("Max pinned cache entries we consider before giving up"),
 
     Option("bluestore_cache_type", Option::TYPE_STR, Option::LEVEL_DEV)