release(commit) - related to EDM pool full

D A

release(commit) - related to EDM pool full
Hi,
This is DB2 v7 for OS/390.
After our stored procedure crashes on EDM pool full I can see that the
connection is still holding a locks on the resources (see below). All our
packages are bound with release(commit). How is it possible that for some
locks their duration is set to ALLOCATION ? Also, after the crash, before
the disconnect I issue manually COMMIT - guess what - all the 10 locks are
still there (even those with duration-COMMIT+1). They are released only
after ROLLBACK/DISCONNECT. BTW, we don't use cursors defined WITH HOLD.
[Answer to some of my questions could help me find out why our EDM pool
gets filled with package tables.]

I would appreciate any help here.
Daniel

Database Type Status Duration Attribute
Object Count Suspended Function
_ DSNDB01 PAGESET IS ALLOCATION L-LOCK
SPT01 2 NO N/A
_ DSNDB01 TABLE IS ALLOCATION L-LOCK
SPTR 2 NO N/A
_ SP000043 SKPT S COMMIT+1 L-LOCK
ZPZB 1 NO N/A
_ 6 PAGESET IS ALLOCATION L-LOCK
371 (SYSSEQ) 1 NO N/A
_ 6 TABLE IS ALLOCATION L-LOCK
372 (SYSSEQUENCES) 1 NO N/A
_ SP000042 SKPT S COMMIT+1 L-LOCK
ZPZB 1 NO N/A
_ SP000831 SKPT S COMMIT+1 L-LOCK
ZPZB 1 NO N/A
_ SP000828 SKPT S COMMIT+1 L-LOCK
ZPZB 1 NO N/A
_ SQLL7D05 SKPT S COMMIT+1 L-LOCK
NULLID 2 NO N/A
_ SQLC2D03 SKPT S COMMIT+1 L-LOCK
NULLID 2 NO N/A



D A

Re: release(commit) - related to EDM pool full
(in response to D A)
Errr,
correction, the locks are not release until DISCONNECT - even ROLLBACK does
not release them. However, package tables (PT) are released from EDM pool
on ROLLBACK - not on COMMIT though.

Thanks again,
Daniel

On Thu, 19 Dec 2002 08:29:41 -0600, Daniel <[login to unmask email]> wrote:

>Hi,
>This is DB2 v7 for OS/390.
>After our stored procedure crashes on EDM pool full I can see that the
>connection is still holding a locks on the resources (see below). All our
>packages are bound with release(commit). How is it possible that for some
>locks their duration is set to ALLOCATION ? Also, after the crash, before
>the disconnect I issue manually COMMIT - guess what - all the 10 locks are
>still there (even those with duration-COMMIT+1). They are released only
>after ROLLBACK/DISCONNECT. BTW, we don't use cursors defined WITH HOLD.
>[Answer to some of my questions could help me find out why our EDM pool
>gets filled with package tables.]
>
>I would appreciate any help here.
>Daniel
>
> Database Type Status Duration Attribute
> Object Count Suspended Function
> _ DSNDB01 PAGESET IS ALLOCATION L-LOCK
> SPT01 2 NO N/A
> _ DSNDB01 TABLE IS ALLOCATION L-LOCK
> SPTR 2 NO N/A
> _ SP000043 SKPT S COMMIT+1 L-LOCK
> ZPZB 1 NO N/A
> _ 6 PAGESET IS ALLOCATION L-LOCK
> 371 (SYSSEQ) 1 NO N/A
> _ 6 TABLE IS ALLOCATION L-LOCK
> 372 (SYSSEQUENCES) 1 NO N/A
> _ SP000042 SKPT S COMMIT+1 L-LOCK
> ZPZB 1 NO N/A
> _ SP000831 SKPT S COMMIT+1 L-LOCK
> ZPZB 1 NO N/A
> _ SP000828 SKPT S COMMIT+1 L-LOCK
> ZPZB 1 NO N/A
> _ SQLL7D05 SKPT S COMMIT+1 L-LOCK
> NULLID 2 NO N/A
> _ SQLC2D03 SKPT S COMMIT+1 L-LOCK
> NULLID 2 NO N/A
>
>
>
the DB2-L webpage at http://listserv.ylassoc.com. The owners of the list
can