summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorXavi Hernandez <xhernandez@redhat.com>2021-03-17 10:59:54 +0100
committerXavi Hernandez <xhernandez@gmail.com>2021-04-09 18:41:58 +0200
commit909a9a1aa10594a7f53d2bc7fa75ca7f5d2297af (patch)
tree44c0452dc758be01415fd9d6db9b25f9430d7c0c
parent1947a59d08d103982cebb8c7caa0da63088cc970 (diff)
downloadglusterfs-909a9a1aa10594a7f53d2bc7fa75ca7f5d2297af.tar.gz
glusterfs-909a9a1aa10594a7f53d2bc7fa75ca7f5d2297af.tar.xz
glusterfs-909a9a1aa10594a7f53d2bc7fa75ca7f5d2297af.zip
extras: disable lookup-optimize in virt and block groups
lookup-optimize doesn't provide any benefit for virtualized environments and gluster-block workloads, but it's known to cause corruption in some cases when sharding is also enabled and the volume is expanded or shrunk. For this reason, we disable lookup-optimize by default on those environments. Fixes: #2253 Change-Id: I25861aa50b335556a995a9c33318dd3afb41bf71 Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
-rw-r--r--extras/group-distributed-virt1
-rw-r--r--extras/group-gluster-block1
-rw-r--r--extras/group-virt.example1
3 files changed, 3 insertions, 0 deletions
diff --git a/extras/group-distributed-virt b/extras/group-distributed-virt
index a960b76c69..6da3de03fa 100644
--- a/extras/group-distributed-virt
+++ b/extras/group-distributed-virt
@@ -8,3 +8,4 @@ user.cifs=off
client.event-threads=4
server.event-threads=4
performance.client-io-threads=on
+cluster.lookup-optimize=off
diff --git a/extras/group-gluster-block b/extras/group-gluster-block
index 1e398019e6..b8d3e8d9e8 100644
--- a/extras/group-gluster-block
+++ b/extras/group-gluster-block
@@ -25,3 +25,4 @@ features.shard-block-size=64MB
user.cifs=off
server.allow-insecure=on
cluster.choose-local=off
+cluster.lookup-optimize=off
diff --git a/extras/group-virt.example b/extras/group-virt.example
index cc37c98a25..04003b7dfb 100644
--- a/extras/group-virt.example
+++ b/extras/group-virt.example
@@ -22,3 +22,4 @@ server.tcp-user-timeout=20
server.keepalive-time=10
server.keepalive-interval=2
server.keepalive-count=5
+cluster.lookup-optimize=off