RE: Funky problem with SQL ALIAS

Well maybe, yes, it's possible, I think.  The Java developer tells me that
the CREATE & DROP are occurring in the "global" or "mainline" part of the
Java code, not in the spun-off threads (that is the gist of what I
remember her telling me yesterday.)

Here's what I can tell you.  This two-week-old process was working fine
until the other day, and it appears to coincide with an inadvertant
startup of a second (duplicate) Java agent.  But both agents were killed
(!), and I still have the situation from the native iSeries side where I
can neither create nor drop the alias.

Tonight, we're going to get up-to-date on the DB & HIPER group PTFs.  I
have asked our admin to try the SQL DROP ALIAS again after the system is
in a restricted state and, failing that, after an IPL but before PTFs are
applied (if he gets to that scenario).

GA

--- Elvis Budimlic <ebudimlic@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> GA
> 
> Could multithreading on Java side be an issue?  Or SMP on the OS/400
> side?  I guess what I'm getting at, is it possible that execution is 
> not always serial? 
> If so, you might have a case where threadOne would create the view,
> threadTwo steps in (is cycled in and has hold of CPU) and tries to
> create it but fails as view already exists (this failure is monitored 
> for), threadOne drops the view and when threadTwo attempts to drop 
> the view it think exists, it fails.  
> Since failure is not monitored on drop view path, you program bombs.
> If this is the case, easy fix would be to monitor for drop view 
> failure, but I would think multithreading issues would raise their 
> ugly head in other ways then.
> 
> Reading back my email it all seems farfetched, but just trying to
> help...
> 
> Elvis


__________________________________
Do you Yahoo!?
New Yahoo! Photos - easier uploading and sharing.
http://photos.yahoo.com/