Re: strange oracle problem - please help!!

Check the alert log on the prod server and the sessions against the prod and test databases. You will be able to see which DB is being used and whether you shut the prod one down.

WCK wrote:

> It is a major us vendor's travel and expense software so I assume they have
> put it to the regressive testing b4 they roll out. Really, the production
> web servers behave as if I shut/start the production database even though i
> did it on the test environment.
> if I have an PROD tnsnames.ora pointing to the production sid prod database
> on a test environment and i did ;
> set oracle_sid=prod
> svrmgr30>shutdown immediate;
> will this shutdown the oracle database instance in the production
> environment even though I execute it in the test envrionment?
>
> Gordon Cartwright wrote in message <3A86B426.42DF1472_at_yahoo.co.uk>...
> >I think you have at least two problems here, but I don't have enough info
 to
> >resolve either one. I'd need much more info about the application, it's
> >configuration and how it's written.
> >
> >Regarding the 'missing' data that turned up 5 days later. Assuming that the
> >application isn't temporarily holding it on the app/web server, the most
 likely
> >reason for this is that the session which writes the data doesn't commit it
> >until some later event (web server shutdown?). If this were to be the case
 you
> >should be able to see the user sessions connected and locks held. If it's
> >in-house code I'd start looking to see what it does, If it's a 3rd party
 app, I
> >would put some session level traces on to see whats happening.
 Non-committed
> >data would also explain why it doesn't appear on yor export.
> >
> >Regarding the prod/dev shutdown issue. What happens? You said the prod box
 is
> >impacted, but you don't say what happens, error messages etc. Please give
 more
> >info.
> >
> >WCK wrote:
> >
> >> Thanks for offering your help. I checked thru and thru on all the
> >> tnsnames.ora, listener.ora, etc.. all to no avail. the server entries in
 the
> >> tnsnames are IP address oriented (i am also the web programmer, system
> >> analyst, web administrator and system administrator .. sheesh).
> >>
> >> Actually, I got one major suspect right now, my oracle 8.0.6.0.0 on NT
 has a
> >> sga size at 100mb but all the connecting profiles have unlimited
 resources
> >> set. At any one time, I got 100+ connections all running at dedicated
 server
> >> modes. Just 3 weeks ago, the db ran into big problems whereby all the web
> >> application
> >> servers had reported errors in running a common entry save stored
 procedure.
> >> Then I got alot of records missings in the db which the users claimed
 that
> >> they have filed in, system reported the records is saved but log out and
 log
> >> in again, the record disappeared. The backups that I did , export, did
 not
> >> capture the saved data, but it seems to appear after 5 days later. I
> >> strongly suspect that I didn't size up the sga enough. I checked at the
 free
> >> memory at the v$sgastat and there's about 18Mb left.. gotta be this....
> >> oracle and vendor both can't help at all and they blamed each other for
 it.
> >>
> >> CK
> >>
> >> Gordon Cartwright wrote in message <3A8328AA.96327B09_at_yahoo.co.uk>...
> >> >Can you post the tnsnames.ora from the web server box and the
 listener.ora
 from
> >> >the prod and dev servers? Also, if the tnsnames.ora does not specify IP
> >> >addresses, but uses machine names can you give details of the name
 resolution
> >> >methods (DNS, WINS etc).
> >> >
> >> >When the system is up an running, have you had a look at the connections
 from
> >> >the web server? Can you see the sessions against database?
> >> >
> >> >WCK wrote:
> >> >
> >> >> yes.. i have checked all possibilities.. tnsnames, odbc, listener,
 etc.
> >> >> all to no avail. my vendor is baffled and oracle support has given up
 on
 my
> >> >> side.
> >> >> what can possibly happen that can cause this type of strange problem?
> >> >>
> >> >> is it possible that the data somehow never get committed throughout
 the
> >> >> entire 4 days?
> >> >>
> >> >> Gordon Cartwright wrote in message <3A810148.13C46DD2_at_yahoo.co.uk>...
> >> >> >You don't say what makes you think the .99 database is affected, or
 where
 the
> >> >> >webserver software runs. Assuming the web server is on a different
 machine,
> >> >> >check the config on that box to see which DB it is connecting to.
> >> >> >
> >> >> >WCK wrote:
> >> >> >
> >> >> >> My oracle 8.0.6 is running all on NT 4.0 SP6a.
> >> >> >>
> >> >> >> my environment is like this;
> >> >> >> 190.1.1.99 - production environment; PROD instance and database.
> >> >> >> 190.1.1.201 - develop environment; PROD instance and database.
> >> >> >>
> >> >> >> when I am doing shutdown on 190.1.1.201 via svrmgr30, it seems to
 affect
 the
> >> >> >> .99. any idea why this happen? i thought the svrmgr30 affect the
 local
> >> >> >> database environment, even the tnsnames.ora on the .201 doesn't
 have
 an
> >> >> >> entry for the PROD production but is referred as 99PROD.
> >> >> >>
> >> >> >> I think it cause a massive connection misdirection as my web server
> >> >> >> applications reported alot of errors and seems to respond to the
 work
 that
> >> >> >> I was doing in the development environment.
> >> >> >>
> >> >> >> How can this be explained? listener? network? spoofing?
> >> >> >>
> >> >> >> Please help. I am out of ideas already.
> >> >> >> abramswee_at_yahoo.com
> >> >> >
> >> >
> >
Received on Mon Feb 12 2001 - 13:10:09 CST