These queries all seem like an awful lot of unnecessary processing. Here's what I use:
SELECT 1+1 -----Original Message----- From: Antonio Fiol Bonnín [mailto:[EMAIL PROTECTED] Sent: Friday, January 09, 2004 1:25 AM To: Tomcat Users List Subject: Re: Microsoft SQL Server validation query For any database server I can think of: CREATE TABLE validation ( dummy char(1) ); INSERT INTO validation VALUES ( 'X' ); -- and make sure you do this only once And then use as a validation query: SELECT * FROM validation; Well. Yes. I am ashamed of reimplementing Oracle's "DUAL", but if you really want to know of a table that will always be there, provide it yourself. And anyway, I mostly agree with Mike's statement about server-specific params in the config files: "Adding one more ... is not too much of a burden". Only if you set the validation query from outside the config files, which is very unlikely, Mike's statement would not strictly apply IMHO. Antonio Fiol Derek Mahar wrote: >Thank you to all of you for your quick replies. It seems that the >connection pool validation query is not specific to any database server >implementation unless the query statement itself is server-specific >(that is, it refers to a special server system database, table, or >function). I like the idea of querying the server for the date, but >I'm not sure that the date function is standard SQL and portable across >all server implementations. Does anyone know otherwise? > >Derek > >-----Original Message----- >From: Peter Lin [mailto:[EMAIL PROTECTED] >Sent: January 8, 2004 10:35 AM >To: Tomcat Users List >Subject: RE: Microsoft SQL Server validation query > > > >in the past I just select the date from sql server. unless you want to >test a specific table, but that has potential performance impact. > >the safe simple query to see if sql server is alive is to just select >the date. > >peter lin > > >Allistair Crossley <[EMAIL PROTECTED]> wrote: >I would not do that because that would return as many 1s as there are >rows in the table. Something like count(*) may not be the most >efficient but it returns just 1 row always. Also with using 1, you >cannot guarantee a row will come back. > > > >Allistair Crossley >__________________________ > >Intranet Senior Developer >New Media Group, QAS Ltd >Telephone: 020 7819 5343 >__________________________ > > >-----Original Message----- >From: Michael Duffy [mailto:[EMAIL PROTECTED] >Sent: 08 January 2004 15:25 >To: Tomcat Users List >Subject: RE: Microsoft SQL Server validation query > > > >Or even "SELECT 1 FROM TABLE". No COUNT overhead, if >any. - MOD > > >--- Allistair Crossley >wrote: > > >>i think you could use anything .. maybe >> >>SELECT COUNT(*) FROM table >> >>The dual table is an oracle dummy table and is quite >>handy, but I think the validation query can just be >>any old select statement that should return true a >>result always. >> >>ADC >> >>-----Original Message----- >>From: Derek Mahar [mailto:[EMAIL PROTECTED] >>Sent: 08 January 2004 15:15 >>To: Tomcat Users List >>Subject: Microsoft SQL Server validation query >> >> >>Does anyone happen to know which validation query I >>should use for >>Microsoft SQL Server? >> >>Derek >> >> >> >> --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]