Changes between Version 198 and Version 199 of Processing

Timestamp:

May 10, 2015, 9:49:08 AM (11 years ago)

Author:

Mark Huber

Comment:

--

Processing

@@ -729 +729 @@
 
 = Node Trouble = 
-Sometimes nodes will crash or become unresponsive.
-
-'''WARNING: currently under no circumstances should the --< stsci >-- nodes be rebooted without Gavin (and?) assistance'''. The raid and processors are separate boxes, doing so in wrong order can do something of a reinit on the raid.
+Sometimes nodes will crash or become unresponsive -- be sure to check console messages and if possibly heavy job use before power cycling.
+
+'''WARNING for the -- stsci nodes --''': consoles and managed PDU's connected to the compute node and can be power cycled like other nodes if becomes unresponsive. The JBOD 
+ JBOD enclsosure requires onsite intervention to power on and after a power outage/facility shutdown, should not be powered up unless confirmation that JBODs are up and running first. If JBOD is unavailable, RAID controller will request for user intervention in order to continue if it does not detect storage volumes, onsite examination of the JBOD will be required in order to continue.
 
 (add console login information)
+
+console --user <username> -v <hostname>
+
+{{{
+^p -- PDU menu
+3 -- off for a minute
+4 -- on
+1 -- exit and monitor boot process (any boot or RAID errors etc)
+}}}
+ 
+ * be sure to log in via console and check mounts home and a data node, date is correct etc
+
+
+'''WARNING''' if node does not reboot, return PDU to off to prevent possible burnout/fire situation (ie do not flip power and walk away). Some nodes take 3-5 minutes to activate boot screen. 
+
 
 Has been cases when ippops2 reboots itself and ganglia does not restart -- in that case