Database Reference
In-Depth Information
LMHB sends a message to LMON to terminate the instances. When LMON is unable to complete the
operation because the machine is busy, these processes get scheduled and the instances do
not die. This delays the entire reconfiguration and causes the system to be in a hung state.
To avoid this, LMHB escalates to a reboot scenario in CSSD in order to create a node eviction.
The sequence of steps is
1. LMHB tries to kill LMON , but does not succeed.
LMON (ospid: 7705) has not moved for 31 sec (1410129684.1410129653)
LMON (ospid: 7705) has not moved for 29 sec (1410129823.1410129794)
LMON (ospid: 7705) has not moved for 28 sec (1410129872.1410129844)
2. LMHB escalates the request to CSSD to kill the node; however, the node kill request
may not be successful because the system is overloaded.
3.
CSSD escalates the priority of the request to evict the machine from the cluster. If
installed CSSD sends request to Intelligent Platform Management Interface (IPMI) to
evict the machine from the cluster.
The next sections will discuss a few of the utilites available for troubleshooting the GI/clusterware environment
is discussed.
Node Verification Using olsnodes
The olsnodes command provides the list of nodes and other information for all nodes participating in the cluster.
Additional cluster-related information could be obtained by adding one or more of the following parameters to
the olsnodes command. To log cluster verification information with more details, options -g (log), -v (verbose),
can be used:
[oracle@ssky3l11p1 ~]$ $GRID_HOME/bin/olsnodes -v -g
lang init : Initializing LXL global
main: Initializing CLSS context
memberlist: No of cluster members configured = 256
memberlist: Allocated mem for lease node vector.
memberlist: Leased NodeList entries used = 1.
memberlist: Getting information for nodenum = 1
memberlist: node_name = ssky3l11p1
memberlist: ctx->lsdata->node_num = 1
print data: Printing the node data
ssky3l11p1
main: olsnodes executed successfully
term: Terminating LSF
[oracle@ssky3l11p1 ~]$
It should be noted that the olsnodes utility could be executed with a combination of the preceding options.
For example, for a summarized view of all the information, it could be executed as shown in the following:
[oracle@ssky3l11p1 ~]$ $GRID_HOME/bin/olsnodes -n -i -g -v
lang init : Initializing LXL global
main: Initializing CLSS context
memberlist: No of cluster members configured = 256
memberlist: Allocated mem for lease node vector.
 
Search WWH ::




Custom Search