host-condition: reduce timeout from 30 to 3 sec
Some new software is on the way to refactor the way OpenBMC firmware
detects if the host is running. Future commits in this series will
elaborate on this.
A long running observation of this existing design is that it never
takes the host 30 seconds to respond if it's actually running. In fact
it's always less then one second if it's actually running because ths
host is much faster then the BMC hardware. The 30s timeout has always
caused confusion because of how long it takes when the host is clearly
not running.
A new phosphor-dbus-interface has moved the interface to determine if
the host is up and running into a simple property read. To implement
that property read, we need to ensure this command returns as quickly as
possible.
This commit reduces the time we give the host to respond to a heartbeat
command from 30 to 3 seconds.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Id288f929a9948da84687dd4455b697263dcf6ac2
diff --git a/configure.ac b/configure.ac
index 2210296..e29b538 100644
--- a/configure.ac
+++ b/configure.ac
@@ -166,7 +166,7 @@
# Timeouts in SECONDS for SoftPowerOff protocol
[AC_ARG_VAR(IPMI_SMS_ATN_ACK_TIMEOUT_SECS, \
[Initial timeout for host to ack and query SMS_ATN from BMC])]
- [AC_DEFINE_UNQUOTED([IPMI_SMS_ATN_ACK_TIMEOUT_SECS], [30], \
+ [AC_DEFINE_UNQUOTED([IPMI_SMS_ATN_ACK_TIMEOUT_SECS], [3], \
[Timeout for host to ack and query SMS_ATN from BMC])]
[AC_ARG_VAR(IPMI_HOST_SHUTDOWN_COMPLETE_TIMEOUT_SECS, [Wait time for host to shutdown])]