Hi Everyone:
=20
We are having a lot of deadlocks on some specific pages. We analize at
the record level what we have, and we have a calc record and via records
on the same area, and in the chains there are more than 100 recors via
for any calc owner.The order in the set is next. There is a lot of
stores of the via records. The chains updated are for differents calc
records , there is not a concentration on only one calc record. I
checked the print space and notice that in the total of 710,000 pages I
have 12 logicaly fullpages for at least one week, that means that we
are not falling on logically full pages periodically. The area is only
25% full in total, there is a lot of space, but there is a lot of
deadlocks pointing on some pages. I pint the pages and its look like
they are o.k. I'm still try to figure out whats going on, I'm thinking
what is better, to run unload/reload with a page reserve big enough, or
on the other hand unload/reload and decrease the page size of the pages.
What can I check to make take the correct decision . Any ideas and how
to proceed or what to analize to detect the problem ?? I also checked
and I don't see any run unit spending a lot of time taking the area, or
any exclusive update use of the area. Any help wil be great!!!
=20
Thanks.
=20
J. Sotela=20
_________________________________________________________________
Discover the new Windows Vista
http://search.msn.com/results.aspx?q=3Dwindows+vista&mkt=3Den-US&form=3DQ=
BRE
"
IDMS 3rd-party providers forum
IDMSVENDOR-L@LISTSERV.IUASSN.COM
SMTP
IDMSVENDOR-L@LISTSERV.IUASSN.COM
IDMSVENDOR-L@LISTSERV.IUASSN.COM
SMTP
Normal
Normal
Re: Deadlocks on a specific pages
"Hi Richard:
=20
About your questions:
=20
The message of the deadlock point to a page=2C not a record=2C then I guess=
the deadlock is with smp.
The record is fixed lenght.The owner record is calc and spread accordingly.=
I check the differents calc records and they are on diffferent pages. Via =
records are located in the same page or in other pages.
The records stores are the via records=2C not the calc records.
=20
Thanks.
=20
J. Sotela
=20
Date: Fri=2C 12 Dec 2008 09:07:48 -0500> From: Richard.Pierce@STATE.MA.US=
Subject: Re: Deadlocks on a specific pages> To: IDMS-L@LISTSERV.IUASSN.CO=
M> > Javier=2C> > The big question is what kind of a deadlock? Space manage=
ment or> individual record deadlock?> > Also=2C are these/this record varia=
ble? > > One other question=2C how distributed are the calc records? Even a=
cross> the area or clustered among some page portion of the area?> > We hav=
e deadlock problems=2C but it's because we have multiple> transactions stor=
ing the same record type with the same calc key. Can't> change so they've p=
rogrammed around the problem.> > Dick> > Richard Pierce> (617) 973-8911> ri=
chard.pierce@state.ma.us> > >
Original Message-----> From: IDMS Public=
Discussion Forum [mailTo:IDMS-L@listserv.iuassn.com]> On Behalf Of javier =
sotela> Sent: Friday December 12=2C 2008 8:46 AM> To: IDMS-L@listserv.iuass=
n.com> Subject: Deadlocks on a specific pages> > > Hi Everyone:> > We are h=
aving a lot of deadlocks on some specific pages. We analize at> the record =
level what we have=2C and we have a calc record and via records> on the sam=
e area=2C and in the chains there are more than 100 recors via> for any cal=
c owner.The order in the set is next. There is a lot of> stores of the via =
records. The chains updated are for differents calc> records =2C there is n=
ot a concentration on only one calc record. I> checked the print space and =
notice that in the total of 710=2C000 pages I> have 12 logicaly fullpages f=
or at least one week=2C that means that we> are not falling on logically fu=
ll pages periodically. The area is only> 25% full in total=2C there is a lo=
t of space=2C but there is a lot of> deadlocks pointing on some pages. I pi=
nt the pages and its look like> they are o.k. I'm still try to figure out w=
hats going on=2C I'm thinking> what is better=2C to run unload/reload with =
a page reserve big enough=2C or> on the other hand unload/reload and decrea=
se the page size of the pages.> What can I check to make take the correct d=
ecision . Any ideas and how> to proceed or what to analize to detect the pr=
oblem ?? I also checked> and I don't see any run unit spending a lot of tim=
e taking the area=2C or> any exclusive update use of the area. Any help wil=
be great!!!> > Thanks.> > J. Sotela > ____________________________________=
_____________________________> Discover the new Windows Vista> http://searc=
h.msn.com/results.aspx?q=3Dwindows+vista&mkt=3Den-US&form=3DQBRE
_________________________________________________________________
Invite your mail contacts to join your friends list with Windows Live Space=
s. It's easy!
http://spaces.live.com/spacesapi.aspx?wx_action=3Dcreate&wx_url=3D/friends.=
aspx&mkt=3Den-us=
"
IDMS 3rd-party providers forum
IDMSVENDOR-L@LISTSERV.IUASSN.COM
SMTP
IDMSVENDOR-L@LISTSERV.IUASSN.COM
IDMSVENDOR-L@LISTSERV.IUASSN.COM
SMTP
Normal
Normal
Re: Deadlocks on a specific pages
"Hi Richard:
About your questions:
The message of the deadlock point to a page, not a record, then I guess the deadlock is with smp.
The record is fixed lenght.The owner record is calc and spread accordingly. I check the differents calc records and they are on diffferent pages. Via records are located in the same page or in other pages.
The records stores are the via records, not the calc records.
Thanks.
J. Sotela
Date: Fri, 12 Dec 2008 09:07:48 -0500> From: Richard.Pierce@STATE.MA.US> Subject: Re: Deadlocks on a specific pages> To: IDMS-L@LISTSERV.IUASSN.COM> > Javier,> > The big question is what kind of a deadlock? Space management or> individual record deadlock?> > Also, are these/this record variable? > > One other question, how distributed are the calc records? Even across> the area or clustered among some page portion of the area?> > We have deadlock problems, but it's because we have multiple> transactions storing the same record type with the same calc key. Can't> change so they've programmed around the problem.> > Dick> > Richard Pierce> (617) 973-8911> richard.pierce@state.ma.us> > >