Hi, all
I revised timeout values(start action and lock_timeout) and long
description of the lock_timeout of the sfex RA.
The old start action timeout value(600s) is too long.
It originally considered the execution time of fsck,
but the sfex RA have no fsck function now.
The new start action timeout value is calculated by the following formula.
start timeout = collision_timeout + lock_timeout + "safety margin"
The "safety margin" is decided within the range of about 10-20 seconds.
# It depends on your system requirement.
The default value of the collision_timeout is 1 second as usual.
The default value of the lock_timeout is calculated by the following
formula.
lock_timeout = monitor_interval + "The expiration time of the lock"
I suggest 90 seconds as a default value of the "The expiration time of
the lock", but users should change it in consideration of access delay
to the shared disk and the switch time of the multipath driver.
Please see attached patch and let me know if you have any comments.
Best Regards.
--
NAKAHIRA Kazutomo
Infrastructure Software Technology Unit
NTT Open Source Software Center
# HG changeset patch
# User NAKAHIRA Kazutomo <nakahira.kazut...@oss.ntt.co.jp>
# Date 1289798355 -32400
# Node ID c3de51abdc0b56e6bfb3457077e3c9a8ccca0d32
# Parent 6296c9f24b2f31153c47bc34eb6c0d2e3e40eab9
sfex RA: revise timeout values
diff -r 6296c9f24b2f -r c3de51abdc0b heartbeat/sfex
--- a/heartbeat/sfex Tue Nov 09 11:46:15 2010 +0100
+++ b/heartbeat/sfex Mon Nov 15 14:19:15 2010 +0900
@@ -92,18 +92,30 @@
</parameter>
<parameter name="lock_timeout" unique="0" required="0">
<longdesc lang="en">
-Valid term of lock(sec). Default is 20 seconds.
+Valid term of lock(sec). Default is 100 seconds.
+The lock_timeout is calculated by the following formula.
+
+ lock_timeout = monitor_interval + "The expiration time of the lock"
+
+We suggest 90 seconds as a default value of the "The expiration time of the lock", but you should change it in consideration of access delay to the shared disk and the switch time of the multipath driver.
+
+The lock timeout have an impact on start action timeout because start action timeout value is calculated by the following formula.
+
+ start timeout = collision_timeout + lock_timeout + "safety margin"
+
+The "safety margin" is decided within the range of about 10-20 seconds(It depends on your system requirement).
</longdesc>
<shortdesc lang="en">Valid term of lock</shortdesc>
-<content type="integer" default="20" />
+<content type="integer" default="100" />
</parameter>
</parameters>
<actions>
-<action name="start" timeout="600" />
+<action name="start" timeout="120s" />
<action name="stop" timeout="20s" />
-<action name="monitor" depth="0" timeout="10" interval="10" />
-<action name="meta-data" timeout="5" />
+<action name="monitor" depth="0" timeout="10s" interval="10s" />
+<action name="meta-data" timeout="5s" />
+<action name="validate-all" timeout="5s" />
</actions>
</resource-agent>
END
@@ -246,7 +258,7 @@
DEVICE=$OCF_RESKEY_device
INDEX=${OCF_RESKEY_index:-1}
COLLISION_TIMEOUT=${OCF_RESKEY_collision_timeout:-1}
-LOCK_TIMEOUT=${OCF_RESKEY_lock_timeout:-20}
+LOCK_TIMEOUT=${OCF_RESKEY_lock_timeout:-100}
MONITOR_INTERVAL=${OCF_RESKEY_monitor_interval:-10}
sfex_validate () {
_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
