Erroneous errors in sccheck output?

Question

Hi All,
I'm using Solaris 10 u6 & Sun Cluster 3.2 u2 to build out a cluster framework.
I'm seeing 2 errors in the sccheck log output and I'm suspicious that they are false positives:
1.
CHECK ID : 1212
SEVERITY : CRITICAL
PROBLEM : Unsupported Real Time processes running on a Sun Cluster node may interfere with the proper operation of the cluster.
ANALYSIS :
A Real Time (RT) process other than a Sun Cluster daemon or an application daemon
qualified to run RT with Sun Cluster was found running on this Sun Cluster node. A Real
Time process that has not been qualified to run with Sun Cluster may interfere with the
proper operation of the cluster and result in unexpected cluster behavior. Processes
/usr/cluster/lib/sc/cl_execd and /usr/cluster/lib/sc/cl_execd were found running as Real
Time.
So when I check to see what is running in the Real Time scheduling class:
# ps -cafe | grep RT | grep -v grep
root 373 371 RT 100 14:32:54 ? 0:00 /usr/cluster/lib/sc/cl_execd
root 394 1 RT 100 14:32:55 ? 0:00 /usr/cluster/lib/sc/clexecd
root 395 394 RT 100 14:32:55 ? 0:00 /usr/cluster/lib/sc/clexecd
root 371 1 RT 100 14:32:54 ? 0:00 /usr/cluster/lib/sc/cl_execd
root 1195 1 RT 100 14:33:26 ? 0:00 /usr/cluster/lib/sc/rpc.pmfd
root 1257 390 RT 101 14:33:28 ? 0:00 /usr/cluster/lib/sc/rgmd -z global
root 1203 1 RT 100 14:33:26 ? 0:00 /usr/cluster/lib/sc/rpc.fed
I see that these are cluster processes, so I wonder are these relevant?
2.
CHECK ID : 1710
SEVERITY : CRITICAL
PROBLEM : Cluster failover/switchover might fail because CCR files are missing or corrupt on a cluster node.
ANALYSIS :
The CCR directory or files in this directory is missing or corrupt. Without these files,
the proper node configuration can not be determined. This node will fail to takeover a
data service in the event that the node mastering the service fails. During a failover,
one of the remaining nodes will need to bring up the data service so that clients can
access that service. However, this is not possible because the configuration parameters
for this service are lost. This can lead to a loss of service.
File cluster/etc/cluster/ccr/dcs_service_classes is missing. File
cluster/etc/cluster/ccr/dcs_service_keys is missing. File
cluster/etc/cluster/ccr/did_types is missing. File cluster/etc/cluster/ccr/did_instances
is missing. File cluster/etc/cluster/ccr/directory is missing. File
cluster/etc/cluster/ccr/epoch is missing. File cluster/etc/cluster/ccr/infrastructure is
missing.
However I believe that this is exlpained by the fact that the ccr directory is was created with a subdirectory called global
# ls /etc/cluster/ccr/global
cz_network_ccr dcs_service_3.bak did_types.bak
dcs_service_1 dcs_service_4 directory
dcs_service_1.bak dcs_service_4.bak directory.bak
dcs_service_10 dcs_service_5 dpm_autoreboot_table
I have no ide where this came from, as I didn't explicitly create the directory or point the installer/config at it when creating the cluster.
Note that the cluster seems to work just fine.
I see the exact same log errors on the cluster a colleague has built, and basically I want to establish whether I can ignore them or not.
Note the clusters appear to be alive and healthy otherwise!
Thanks for comments,
Paul.

Answer 1

I've just got this response from the person who knows about sccheck and cluster check. He says:
6813854 Check 1710 in sccheck broken under SC3.2U2
as observer noted: global subdir was added: ignore violation
6813853 Check 1212 in sccheck broken under SC3.2U2
ignore complaints about /usr/cluster/lib/sc/cl_execd: is valid process
Others:
6813852 Check 1872 in sccheck broken under SC3.2U2
I think this is same cause as 1710
6813855 Check 3067 in sccheck broken under SC3.2U2: x86 only
ignore complaint about OBP under x86
6813856 Check 2375 in sccheck broken under SC3.2U2
ignore violation if using lofi
Should be fixed in next patch set. I can't give you a date for that though.
Tim
---