TRANSFER Release D30 Documentation Supplement
Improved TMANAGER HELD Reads
1-10 118474—TRANSFER Release D30 Documentation Supplement
TRANSFER D30 Documentation Supplement
Improved TMANAGER HELD Reads
The reading of HELD records required a read through the entire ready file, which made
it difficult to check for held records on a system with a large backlog of tasks to
perform.
When a message is held, its priority is also altered to move it to the bottom of the queue.
SMSERV now uses the new priority (bit 0 = 1) to keyposition into the READY file.
TSAMPLE Priority 0 Counting
Two new parameters are available in TSAMPLE to indicate that only non-priority 0
tasks are counted during peak usage. These parameters allow samples to be taken of
high priority work without having to read all the priority 0 records.
PEAKSTARTHOUR hour
Where hour indicates that any sample taken after this hour but before
PEAKENDHOUR does not count priority 0 tasks.
PEAKENDHOUR hour
Where hour ends peak time. Samples taken after this hour but before
PEAKSTARTHOUR count all records.
An asterisk (*) is used to mark samples that do not include priority 0 records.
External Object Transport Enhancements
The transport mechanism of external objects has been made more robust. Previously, if
the item descriptor on the receiving end was found, the assumption was that the file
existed and transport could possibly fail later. This area of code was examined and made
more resistant to external problems.
•
TWORK reports the file name and operating system error with any 4119 error.
•
TRECV reports the parent ID, object ID and file name for any 4118 error.
•
TRECV checks, at get-extobj-time, if the file is there. If not, it tells TWORK to
duplicate the file again.
Text Server Utilization Improvement
Servers rely on the Text Server to initialize their context's language and character set.
Initialization is required when an incoming request requires a context with a different
language or character set from the previous request but is not required if these are the
same. The check for matching language and character sets is improved.
The Text Server received more requests than necessary; these calls have been eliminated.
The Text Server normally consists of a single process, which can be a bottleneck on a
large system without this improvement.