- Master Node Check
- Selected Node Check
When the sequencer of an active operation terminates unexpectedly, an operation transitions into a recoverable state. This is most common when a reboot is required within an operation. Once a recoverable-state operation is recovered, it becomes active again and is treated as an active operation. Recoverable-state operations require special handling when they are detected by the Selected Nodes Check.
Fatal Error Detection
When fatal errors are detected, they are listed on the summary screen as fatal. You cannot proceed with your current operation until all fatal errors have been resolved. Fatal errors are most commonly caused by a failure to contact the port manager named in the locked file or an internal PUT error.
When the port manager cannot be contacted, the Simultaneous Operation check cannot determine the state of the operation indicated in the locked file. This is encountered when a locked file from a previous operation exists on a node, but the master node that created the locked file has been disconnected or powered down. Most of the time, this problem is corrected by the expiration of old locked files. If manually removing the locked file does not correct this problem, the problem could be internal to PUT. Check the sequencer logs in the applicable directory for additional information on the failure:
/var/opt/teradata/TDput/fileservice/logs
Master Node Check
When an operation is active or “In Progress” on a master node, PUT does not allow you to start another operation until the current one is no longer active. Attempting to start another operation results in a warning message and redisplays the main screen after you click OK. This check is referred to as the master node check and is performed on the master node before starting any operation. Master nodes are allowed to have recoverable and active operations at the same time, as long as only one is active. Any attempt to recover a recoverable-state operation while another operation is active results in the same warning message described above.
Selected Node Check
Locked files expire one hour after they are created. When the port manager indicates a locked file cannot be contacted and the locked file is expired, the new operation automatically deletes the old locked file and continues. If the locked file has not expired, this is considered an error and the new operation cannot proceed. It is possible to get around this type of error by manually deleting the locked files on the nodes that failed. The SimOpLock file that must be deleted is located at the appropriate location: /var/opt/teradata/TDput/fileservices/.
This should be used as a last resort, since you are avoiding the simultaneous operation protection provided by PUT. If the operation that created the locked file is still active when the file is deleted, unpredictable results can occur by starting another operation. If possible, it is safer to wait until the old locked file expires.
Recoverable Operations
It is possible for an operation to be recovered between the time that your current operation detects it and when you attempt to cancel it. If this happens, the Simultaneous Operation check detects that the recovered operation is now active and does not allow you to cancel it and proceed with a new operation. Once an operation is recovered, it is active again and the selected node check detects it and prevents subsequent operations from selecting those nodes.
Recoverable-state operations are allowed on a master node while another operation is active, as long as none of the same nodes are selected for the two operations.