Re: Record lock, no retry?

[Booth Martin <booth@xxxxxxxxxxxx>]

My experience is that usually it is one or two programs where this is an 
issue and fixing those programs makes the problem go away.

First, why is a record locked for more than a fraction of a second, 
ever?  Almost always it is a maintenance program where a record is read 
in as  locked, and then displayed on a user's screen waiting for 
user-action. Stop doing that and record locks disappear,

 * read the record (unlocked).
 * display the record, along with a hidden data structure of the
   original values.
 * when ENTER pressed, check new values to original values.
 * If changes were made, get the record again (locked).
 * Compare new values to original values (maybe changed by another user?)
 * if OK, do the update.  Record lock is sub-second.



On 9/6/2018 12:33 PM, Justin Taylor wrote:
> When I convinced the boss that an OS retry option wasn't practical, I implemented something like you suggest.  We already have an old program (that I prob found somewhere) that you pass it a PF name and an RRN, and it will send the current job a break msg saying who has that row locked.  All I had to do was monitor for the lock, do a no-lock I/O to get the RRN and then call this pgm.
>
> Thanks everyone!