HP XC System Software Administration Guide Version 3.2

21.1.2 Mismatched Secure Shell Keys
If a node on your system has a mismatched Secure Shell (ssh) key, review the following list for
the source of the problem:
• The node was not imaged, and was booted an old image, which had older ssh keys. In this
instance, it is the image, not the keys, that is out of synchronization.
You can solve this problem by imaging the node properly and rebooting.
• The keys were regenerated on the head node. Typically, the cluster_config utlity was
run with the option to regenerate the keys.
If you absolutely must synchronize keys, enter the following command:
# /opt/hptc/nagios/libexec/check_keys --update
This command uses the Nagios nrpe plug-in executor to create a checksum for the keys
with the superuser. If a node has a key is out of synchronization, the node requests and
receives a new copy of the key.
This method offers a means of recovery if the root ssh keys are somehow damaged or
corrupted.
21.1.3 NFS Mount Failure (Permission Denied)
It is possible to experience a mount failure when nodes image, boot, and attempt to NFS mount
the /hptc_cluster file system. The nodes receive a Permission denied error message from
the attempt to mount the /hptc_cluster file system.
If nodes receive a Permission denied error, run the following commands on the head node
to restart nfs and re-run node configuration scripts and restart services on all other nodes:
# service nfs restart
# pdsh -a touch /var/hptc/nconfig.1st
# stopsys
# startsys
21.1.4 NFS Attribute Caching on Large-Scale Systems
On large-scale HP XC systems, NFS attribute caching can cause inconsistencies when large
numbers of client nodes are asked to access files that have been updated recently and the attribute
caching data on the client has not expired. Disabling attribute caching eliminates the problem
but may degrade performance for frequent file operations requiring file attribute data.
You can disable attribute caching using the noac mount option on all external NFS mounts.
21.1.5 Stale Metrics Data
On systems with the administration network on a 10.x.x.x subnet, it is possible that the database
permissions are not set correctly for logging metrics data. The symptom is that several Nagios
services, such as NodeInfo and System Free Space report WARNING: Data is stale alerts.
Enter the following command to determine whether this is occurring:
# /opt/hptc/nagios/libexec/check_metrics -d 0 btime
If the command output reports an error similar to Cannot connect to server: Access
denied for user 'logger', it is likely that the database permissions are not set correctly.
A patch, hptc-cluconfig-4-40.2sp, fixes this problem, but installing the patch requires you
to reconfigure and reimage the cluster.
To avoid reconfiguring and reimaging the cluster, contact the HP XC support team at
[e-mail] and request the fixpriv script. This script updates the permissions in
the MySQL database.
After running the script, run the check_metrics command again to confirm the fix.
246 Troubleshooting