Reading a record without placing a lock

Dear All,
I have one server application running which continously reading and
updating a DB.
While there is a Reporting tool which generating reports.
The reports can fail, but the server application cannot. So, I need to
run a query in the Reporting tool without placing a lock on the DB
(totally transaprent to the server).
Currently this is what I am doing.
SELECT * FROM Data WITH (NOLOCK);
Question:
1. Am I doing it correctly?
2. Is there a better way of doing? Example: setting the LOCK MODE
(instead of specifying NOLOCK on every command)
Thanks in advance.<ckkwan@.my-deja.com> wrote in message
news:e9061fce-a45f-4866-9f90-9a3e0043c5fc@.s33g2000pri.googlegroups.com...
> Dear All,
> I have one server application running which continously reading and
> updating a DB.
> While there is a Reporting tool which generating reports.
> The reports can fail, but the server application cannot. So, I need to
> run a query in the Reporting tool without placing a lock on the DB
> (totally transaprent to the server).
> Currently this is what I am doing.
> SELECT * FROM Data WITH (NOLOCK);
> Question:
> 1. Am I doing it correctly?
> 2. Is there a better way of doing? Example: setting the LOCK MODE
> (instead of specifying NOLOCK on every command)
> Thanks in advance.
Hi
You can set the transaction isolation level to read uncommitted for the
session, but then you are potentially going to have dirty reads. Other ways
to do this would be to offload the reporting database either by using log
shipping, replication, mirroring or a snapshot.
John|||Hi
> 1. Am I doing it correctly?
No. What if some user inserts/deletes the row while you are reading. You
are about to get an inconsistent data. For example
you have tree pages with data like a) 10,40,60 b) 80,100,90 c)110,70,85 ,
so while you read page A another connection inserst the value let me say
50, but you have already read data from the page A , so it moves the all
data to a new created page so now that data looks like a)10,40 ,50,
b) 80,100,90 c)110,70,85 ,d)60 ... and as you keep reading you get
60(duplicate) from page D as well.
> 2. Is there a better way of doing? Example: setting the LOCK MODE
> (instead of specifying NOLOCK on every command)
Yes , you can use TABLOCK hint or if you use SQL Server 2005 take a look
at SNAPSHOT ISOLATION LEVEL in the BOL
<ckkwan@.my-deja.com> wrote in message
news:e9061fce-a45f-4866-9f90-9a3e0043c5fc@.s33g2000pri.googlegroups.com...
> Dear All,
> I have one server application running which continously reading and
> updating a DB.
> While there is a Reporting tool which generating reports.
> The reports can fail, but the server application cannot. So, I need to
> run a query in the Reporting tool without placing a lock on the DB
> (totally transaprent to the server).
> Currently this is what I am doing.
> SELECT * FROM Data WITH (NOLOCK);
> Question:
> 1. Am I doing it correctly?
> 2. Is there a better way of doing? Example: setting the LOCK MODE
> (instead of specifying NOLOCK on every command)
> Thanks in advance.|||Thanks for the info, as I have mentioned earlier in my post, the
Reporting tool can afford to fail, so I don't really mind the data
inconsistency.
There is something like LOCK MODE in informix where we can set the
LOCK hint globally for a specific connection. Is there something
similar in SqlServer (and no, this is not the ISOLATION LEVEL).
On Apr 13, 7:24=A0pm, "Uri Dimant" <u...@.iscar.co.il> wrote:
> Hi
> > 1. Am I doing it correctly?
> No. What if some user inserts/deletes the row =A0while =A0you are reading.= You
> are about to get an inconsistent data. For example
> =A0you have tree pages with data =A0like a) 10,40,60 b) 80,100,90 =A0c)110=,70,85 ,
> so while =A0you read page A =A0another connection inserst the value let me= say
> 50, but you have already read data =A0from the page A , so it moves the al=l
> data to a new created page so now that data looks like =A0a)10,40 ,50,
> b) 80,100,90 =A0c)110,70,85 ,d)60 ... and =A0as you keep reading =A0you g=et
> 60(duplicate) from page D as well.
> > 2. Is there a better way of doing? Example: setting the LOCK MODE
> > (instead of specifying NOLOCK on every command)
> Yes , you can use TABLOCK hint =A0or if you use SQL Server 2005 =A0take a =look
> at SNAPSHOT ISOLATION LEVEL in the BOL
> <ckk...@.my-deja.com> wrote in message|||<ckkwan@.my-deja.com> wrote in message
news:2d0f36bc-e70e-40c6-8961-145d20c66e59@.q1g2000prf.googlegroups.com...
Thanks for the info, as I have mentioned earlier in my post, the
Reporting tool can afford to fail, so I don't really mind the data
inconsistency.
There is something like LOCK MODE in informix where we can set the
LOCK hint globally for a specific connection. Is there something
similar in SqlServer (and no, this is not the ISOLATION LEVEL).
On Apr 13, 7:24 pm, "Uri Dimant" <u...@.iscar.co.il> wrote:
> Hi
> > 1. Am I doing it correctly?
> No. What if some user inserts/deletes the row while you are reading. You
> are about to get an inconsistent data. For example
> you have tree pages with data like a) 10,40,60 b) 80,100,90 c)110,70,85 ,
> so while you read page A another connection inserst the value let me say
> 50, but you have already read data from the page A , so it moves the all
> data to a new created page so now that data looks like a)10,40 ,50,
> b) 80,100,90 c)110,70,85 ,d)60 ... and as you keep reading you get
> 60(duplicate) from page D as well.
> > 2. Is there a better way of doing? Example: setting the LOCK MODE
> > (instead of specifying NOLOCK on every command)
> Yes , you can use TABLOCK hint or if you use SQL Server 2005 take a look
> at SNAPSHOT ISOLATION LEVEL in the BOL
> <ckk...@.my-deja.com> wrote in message
Hi
I can't see how unreliability and inconsistence made the user requirements
for this system!
John|||<<There is something like LOCK MODE in informix where we can set the
LOCK hint globally for a specific connection. Is there something
similar in SqlServer (and no, this is not the ISOLATION LEVEL).>>
The ANSI SQL Compliant way to describe how much you want to be isolated from other users is the SET
TRANSACTION ISOLATION command. SQL Server supports this, and READ UNCOMMITTED seems to do what you
want. Apparently Informix has a non-standard command named LOCK MODE, something that SQL Server do
not have. Assuming these indeed do the same thing, I support MS for using the ANSI SQL compliant
name for the command instead of some other command name. If they do not do the same, perhaps you can
enlighten un in what way they differ?
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
http://sqlblog.com/blogs/tibor_karaszi
<ckkwan@.my-deja.com> wrote in message
news:2d0f36bc-e70e-40c6-8961-145d20c66e59@.q1g2000prf.googlegroups.com...
Thanks for the info, as I have mentioned earlier in my post, the
Reporting tool can afford to fail, so I don't really mind the data
inconsistency.
There is something like LOCK MODE in informix where we can set the
LOCK hint globally for a specific connection. Is there something
similar in SqlServer (and no, this is not the ISOLATION LEVEL).
On Apr 13, 7:24 pm, "Uri Dimant" <u...@.iscar.co.il> wrote:
> Hi
> > 1. Am I doing it correctly?
> No. What if some user inserts/deletes the row while you are reading. You
> are about to get an inconsistent data. For example
> you have tree pages with data like a) 10,40,60 b) 80,100,90 c)110,70,85 ,
> so while you read page A another connection inserst the value let me say
> 50, but you have already read data from the page A , so it moves the all
> data to a new created page so now that data looks like a)10,40 ,50,
> b) 80,100,90 c)110,70,85 ,d)60 ... and as you keep reading you get
> 60(duplicate) from page D as well.
> > 2. Is there a better way of doing? Example: setting the LOCK MODE
> > (instead of specifying NOLOCK on every command)
> Yes , you can use TABLOCK hint or if you use SQL Server 2005 take a look
> at SNAPSHOT ISOLATION LEVEL in the BOL
> <ckk...@.my-deja.com> wrote in message

RDA problem

I am attempting to execute an RDA.pull from my master database and it is generating exceptions the second time I attempt the pull.

NOTE: I always drop the table prior to executing the pull.

It seems that after the first time I execute a pull against a particular table name on the PDA I get the following:

"A duplicate value cannot be inserted into a unique index. [ Table name = __sysRDASubscriptions,Constraint name = c_LocalTableName ]"

Given that the table doesn't exist when I do the pull (I have already dropped it, and verified it's non-existence via the Query Analyzer), why does this error appear? It's almost as though some artifact of the table still exists in the background somewhere and it doesn't like responding to a pull...

Anyone see this before?

This only seems to manifest when I choose one of the trackingOn options--when tracking is off, this doesn't happen.|||

If you call CompactDatabase after dropping the table, does that resolve the error ? This is just a troubleshooting step to try.

Thank you!

Syed N. Yousuf

Microsoft Developer Support Professional

This posting is provided “AS IS” with no warranties, and confers no rights.

|||

Hmm..Drop all these

1) The table you have pulled

2) The error table (if any you chose)

3) The PK Indexes of the pulled table (if you pulled with IndexesOn)

4) The Constraints for that table

Thanks,

Laxmi Narsimha Rao ORUGANTI, MSFT, SQL Mobile, Microsoft Corporation

|||

Good idea Syed, but no luck. It didn't seem to make any difference if I compacted the database and then pulled into the new db created.

-Kevin

|||

Thanks for your suggestions Laxmi

I'm having trouble with this. How can I drop the constraints and PK Indexes if I don't know what they're called? It would seem that SQL Server Mobile has assigned generated names to these, and I can't figure out how to drop them.

Is this even possible?

-Kevin

|||

To know the indexes and their information such as on what table and what columns they are created use this query:

SELECT * FROM INFORMATION_SCHEMA.INDEXES;

Similarly for constraints,

SELECT * FROM INFORMATION_SCHEMA.TABLE_CONSTRAINTS;

SELECT * FROM INFORMATION_SCHEMA.REFERENTIAL_CONSTRAINTS

Hope this helps!

Thanks,

Laxmi Narsimha Rao ORUGANTI, MSFT, SQL Mobile, Microsoft Corporation

|||

Hi Laxmi

Thanks for the info. This allows me to see the constraints and indexes, but Sql Mobile doesn't seem to keen on giving them up. When I try to delete my PK constraint (using ALTER TABLE xxx DROP CONSTRAINT) I get a message saying "DDL Operations on this table are restricted".

One thing I found that works is to simply delete the database file and recreate it prior to calling rda.pull. This would seem to be the simplest--are there any problems you see in this approach?

-Kevin

|||

No problem, only if there is no other table in that database. Please make sure that you transfer the data from old database to new database in this approach.

Thanks,

Laxmi Narsimha Rao ORUGANTI, MSFT, SQL Mobile, Microsoft Corporation