ca.portal.admin

Re:Re: [IDMSVENDOR-L] Access to Lock Data Used in PMRM?

Discussion created by ca.portal.admin on Mar 10, 2010
OK to jump back in...
Yes they have CAS, yes they have modified the code so I cannot tell what
is
ust older CAS code and what is modified.
Since I don't know yet which particular programs are holding the
ecords/locks, I was looking around at some potential candidates and
earched through ALL of the dialogs looking
or EXCLUSIVE. I found that a bunch of programs are using TSTLCK95 - and
ome are using a varient of it.
rom what I've seen this is called in most cases only in the RESPONSE
when
reparing to UPDATE.
STLCK does a TEST RETURN NOTIFICATION and KEEP LONGTERM 'A' UPGRADE
XCLUSIVE.
*I do see that most of the programs I have looked at do a KEEP LONGTERM
ALL
ELEASE before EXEC NEXT FUNCTION, but since I don't know which
program(s)
o concentrate on, I do not know that all of the programs are doing a
elease before continuing on (in every case).
*Actually I am pretty sure that some programs are NOT releasing locks-
even
hen no processing is going on if I look in LOCKMON or OPER I see a bunch
of
sers with dozens of Notify locks.
However, the STALL victim that I am looking into seems to be hanging up
in
he premap (which is also readied as UPDATE)
I see that the premap does do a KEEP LONGTERM ALL RELEASE - but I have
ever been totally clear on longterm locks - I assume that release only
pplies to LT locks set within the same program?)
Don - I believe that you wrote that LONGTERM locks do not lock the
records
contributing to stalls)? Would that still apply if there is an upgade
to
xclusive?

ROM TSTLCK95:
KEEP LONGTERM 'A' TEST
RETURN NOTIFICATION INTO CSR-RETURN-LOCATION.
F ERROR-STATUS EQ 5121
THEN NEXT COMMAND.
LSE
F DB-ANY-ERROR CALL DBERR99.
LSE
F CSR-RETURN-LOCATION = 2 OR
CSR-RETURN-LOCATION = 3 OR
CSR-RETURN-LOCATION > 5 DO
PROTECT ALL EXCEPT
(AGR-MODE
,AGR-MAP-RESPONSE
,AGR-PASSED-DATA).
MODIFY MAP CURSOR AT FIELD AGR-MAP-RESPONSE.
KEEP LONGTERM ALL RELEASE.
ROLLBACK.
DISPLAY MESSAGE CODE IS 799998. !*REC IN USE-TRY LATER
END.
EEP LONGTERM 'A' UPGRADE EXCLUSIVE.




----Original Message-----
From: Don Casey <donjcasey@COMCAST.NET>
To: IDMS-L@LISTSERV.IUASSN.COM
Sent: Tue, Mar 9, 2010 10:43 pm
Subject: Re: [IDMSVENDOR-L] Access to Lock Data Used in PMRM?

or completeness' sake: the implicit locks placed by default are NOTIFY
cks, and will not and cannot cause a deadlock or stall condition.
Unless
e code somewhere does a KEEP... LONGTERM all potentially problematic
locks
ll be released at the end of the task. I believe Linda has already
spoken
this issue.
nyway; a brute force approach would be to automate a series of DCMT
splays (D AC TA, D RU, D CV, etc) to run every minute or so on a PC
using
tool that can submit them every minute or so and record the output
(e.g.,
mething like Winrunner/Loadrunner). You might be able to correlate the
alls with what's running at the time they occur.
he other, uglier possibility is to use JREPORT8 to fish around in the
urnals looking for signs of a run-unit modifying the DBKEY your rununit
alled on.
on Casey
incipal Consultant
n Right, LLC
----Original Message-----
From: IDMS Public Discussion Forum [mailTo:IDMS-L@LISTSERV.IUASSN.COM] On
half Of Cherlet, Gary (JTS)
Sent: Tuesday, March 09, 2010 6:41 PM
: IDMS-L@LISTSERV.IUASSN.COM
Subject: Re: [IDMSVENDOR-L] Access to Lock Data Used in PMRM?
hris is correct - these are ""long term locks"" which can be held across a
eudo-converse (when there is no active task), as contrasted to Database
cks which can only be held during an active database run unit (which can
t exist when there is no active task).
n ADS there are ""implicit locks"" placed automatically by ADS (there are
ys to manage this through sysgen and compile options), and explicit long
rm locks that can be placed by application code. During the
eudo-converse the long term resources are transferred from the active
task
CE) resource chain to the LTERM resource chain.
TH - cheers - Gary
ary Cherlet
stice Technology Services
partment of Justice, SA Government
lephone +61 (0)8 8226 5199
csimile +61 (0)8 8226 5311
bile +61 (0)41 333 1613
ilTo:gary.cherlet@sa.gov.au
his e-mail message and any attachments are qualified as follows:
dressing: If you have received this e-mail in error, please advise by
ply e-mail to the sender. Please also destroy the original transmission
d its contents. Confidentiality: This e-mail may contain confidential
formation which also may be legally privileged. Only the intended
cipient(s) may access, use, distribute or copy this e-mail. Individual
ews: Unless otherwise indicated, the views expressed are those of the
nder, not Justice Technology Services. Computer Viruses: It is the
cipient's responsibility to check the e-mail and any attached files for
ruses.
----Original Message-----
From: IDMS Public Discussion Forum [mailTo:IDMS-L@LISTSERV.IUASSN.COM] On
half Of Chris Hoelscher
Sent: Wednesday, 10 March 2010 11:58
: IDMS-L@LISTSERV.IUASSN.COM
Subject: Re: [IDMSVENDOR-L] Access to Lock Data Used in PMRM?
ot to muddy the waters further - but I believe that in some cases - a
user
n walk away without logging off (but NOT in the middle of a
ansaction) and STILL hold locks on data resources - in this case i think
e holder would be the LTERM - not a program
hris Hoelscher
MS/DB2 Database Architect
mana Inc
2-476-2538
oelscher@humana.com
ou only need to test the programs that you want to work correctly

From:
nthia Kline <cakask@AOL.COM>
:
MSVENDOR-L@LISTSERV.IUASSN.COM
te:
/09/2010 07:57 PM
Subject:
: [IDMSVENDOR-L] Access to Lock Data Used in PMRM?
nt by:
MS 3rd-party providers forum <IDMSVENDOR-L@LISTSERV.IUASSN.COM>
I have been offloading Logs - I can work with it either way.=20
hat I am trying to determine is:=20
Which program is holding onto the records and causing a stall in
anot=
r program. =20
(since I have been able to execute a TRACE while a STALL is
happening=
d see the TRACE stop right before an OBTAIN OWNER,=20
I figured if I can determine which program is holding onto (loc=
ng) the record the Stalled program needs, then I can check out THAT
prog=
m to find out why it won't let go.)=20
hanks,=20
ndy Kline

Outcomes