Locking problem SqlBase 11.0.2.

forum.gupta.bugreport (2005-2010)
NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 17 Dec 2007, 15:02

 Posted by:  Rein van Erkel 

Hello,

We are in the process of migrating our application from SqlBase 8.5 to
SqlBase 11.0.2 and are experiencing some strange locking problems. I started
a database audit to isolate the conflicting statements bud the contents of
this audit file puzzled me. Not only is the lock type of the first statement
"Unknown" bud more important is the fact that the two statements, that
caused the locking time out, are executed on different databases. The
problem however does not arise if the order by of the second statement is
ommited. I hope someone can shed light on this problem.

Rein van Erkel.

000,17-DEC-07 14:18:25.92,GLOBAL,U4_LOCK,8

008,17-DEC-07 14:19:44.65,LOCK
HOLDER,MVLMAIN,SYSADM,Win32User,RL,-1,0,0,Unknown,

SELECT COUNT(*) FROM LICENTIEGEGEVENS

008,17-DEC-07
14:19:44.65,TIMEOUT,MVLSYST,SYSADM,Win32User,RL,10,1,2304,S-Lock,

SELECT * FROM WF_CA_PARAM WHERE WF_CA_PARAM.CDCONACT = 'ACT' AND
WF_CA_PARAM.TAAL = 'NL' ORDER BY WF_CA_PARAM.PARAMNR

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 18 Dec 2007, 07:29

 Posted by:  F R Bhote 

I have noticed that in SB 11 time sharing is quite a major problem compared
to earlier versions of SB. If you increase the timeout value from 10 you
should get a result - I keep 60 for this reason.

The Unknown has me puzzled too but apparently it comes only on a select.

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 18 Dec 2007, 09:14

 Posted by:  Rein van Erkel 

The timeout is normally set to infinite (-1) but I have reduced it to 10
sec. so I could capture the timeout error in a audit file. We stumbled
across this error because some processes waited endlessly while no
transactions were active. The database (MVLSYST) where the lock timeout
occurred is a database containing fixed information no transactions are ever
performed against that database and we always use the isolation level "RL".
So in my opinion this error should never occur.

Rein.

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 18 Dec 2007, 10:17

 Posted by:  F R Bhote 

Sorry, I missed the - and thought that it was 1.

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 18 Dec 2007, 14:07

 Posted by:  F R Bhote 

By the way, have you set readonly=0?

The default is ON and causes great problems.

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 18 Dec 2007, 17:21

 Posted by:  Rein van Erkel 

I tried both (readonly=0 and readonly=1) no change.

I'm now trying to change the isolation level to 'RC' to see if that improves
the situation.

NewsgroupServer
Robot
Robot
Posts: 116488
Joined: 24 Feb 2017, 12:00
Location: World wide

Locking problem SqlBase 11.0.2.

Post by NewsgroupServer » 20 Dec 2007, 17:08

 Posted by:  Rainer Ebert 

Feroz,

can you please explain the problems that occure with readonly=1.

At the moment I'm evaluating SB 11.0.2 TE. I've turned readonly off because
we only use RL. But I want to change to RC and than I have to set readonly
on.

Regards,
Rainer

Return to “gupta.bugreport”

Who is online

Users browsing this forum: [Ccbot] and 0 guests