Disk unit full considerations

While space is being reserved or data is being written to compressed disk units, it is possible for a unit to become temporarily full.

The storage subsystem controller will detect this situation and attempt to reposition the data on the unit in order to maximize the amount of data that can be stored on the unit. This repositioning of the data increases the effective size of the disk unit. If the storage subsystem controller cannot immediately service a system request, a failure will be returned to the system. The system bases its response to this failure on the type of input/output (I/O) operation that is requested. One of the following scenarios might occur:

  • The system responds to this failure by overflowing the auxiliary storage pool (ASP). Once the overflow occurs, this I/O request will be performed in the system ASP and will succeed.
  • The system responds to this failure by displaying an A6xx 0277 system reference code (SRC) on the control panel in the system unit. The system displays this attention SRC until storage space becomes available on the disk unit that is denoted in the attention SRC.
When the system displays an A6xx 0277 attention SRC on the control panel, it also logs a corresponding A6xx 0277 record in the Product Activity Log. This occurs each time this disk unit full condition is detected. The system also sends message CPI116C 
Compressed disk unit &1 is full
to the QSYSOPR message queue. The system will reissue the failed I/O operation and will continue to display the attention SRC on the system or logical partition control panel until the condition is corrected. When the storage subsystem controller creates sufficient space on the compressed unit to contain the system request, the I/O operation completes successfully and the system resumes normal processing.

While this attention SRC is being displayed, some I/O operations to the affected compressed disk unit might be suspended. As a result, you might observe that jobs which issue I/O operations to the affected unit appear to hang.

To reduce the likelihood of system operations hanging while the storage subsystem recovers from a disk unit full condition, it is suggested that ASPs with compressed units operate with a storage threshold of less than or equal to 90%.

As space on the disk unit continues to be used, eventually the storage subsystem controller can no longer store more data on the unit. At this point, the storage subsystem controller will return a failure on any system requests requiring storage space.