Your Browser is not longer supported

Please use Google Chrome, Mozilla Firefox or Microsoft Edge to view the page correctly
Loading...

{{viewport.spaceProperty.prod}}

Executing the online save

When CREATE-JOBS is used, FDDRL initially checks whether all the requirements for executing the online save have been satisfied.

When START-JOBS is used the main task starts all the subtasks required. In addition, FDDRL starts its own monitoring task (Online Save Control Task = OLSC task) with the job name $FBDOLSC, and this monitors the time limit.
Each subtask requests the resources (disk devices, tape devices and initial volumes). If the home pubset is set for DEFAULT-PUBSET (default setting), the SYSOUT log is disabled. The subtask then waits for the write lock to be set by the main task.

The main task monitors the startup of the subtasks:

  • If a subtask does not reach the synchronization point for setting the write lock, the run is aborted (e.g. insufficient resources).

  • When all the subtasks have reached the synchronization point, the main task sets the write lock for all disks of the home pubset.

Each subtask begins by performing the save. After a disk has been saved, the subtask cancels the write lock for precisely this disk.

During the save the main task monitors the progress of the save operations for the subtasks and ensures that the time limit is observed. The main task executes no outputs while the write lock applies. Nor is the status log created during the online save.
It is not possible to interrupt the main task with K2 (BREAK). When the main task is aborted using the /CANCEL-JOB command, FDDRL cancels the write lock and terminates all subtasks which are still running.

The monitoring task also ensures that the time limit is observed. When the monitoring task is aborted, the write lock is canceled. However, the save jobs continue to run.

As soon as all save jobs have been completed, any write lock which still exists for the disks of the pubset is canceled. The subtask for monitoring the time limit is terminated.

Handling blockades

An FDDRL task is blocked when an unexpected event (e.g. program error with system dump) triggers a write I/O on the home pubset during the write lock. The following measures cause FDDRL to recognize a blockade and to cancel the write lock for the home pubset:

  • The FDDRL main task monitors the progress of the save operations for the subtasks. If no progress can be detected for one minute, the write lock is canceled prematurely.

  • The FDDRL main task and the monitoring task monitor compliance with the time limit specified for the save (TIME-LIMIT operand, default: 5 minutes). The write lock is canceled at the latest when the time limit is reached. However, the online save is continued to the end so that the user can ascertain the scope and duration of the save. The save itself is, however, inconsistent and unusable.

Restrictions

During the write lock all tasks throughout the system which execute write operations on the home pubset are placed in a wait state.

Assumed read operations in a file also trigger a write operation in the catalog when Close is called. Consequently these tasks, too, must wait for the end of the write lock.

It is not possible to start a subsystem during the write lock.

The high I/O load caused by FDDRL obstructs all read accesses to the home pubset.

Administration of the POSIX file systems is not possible during an online save. Write accesses to POSIX file systems are suspended and only executed after the write lock has been reset.