EscVector schreef:
> Frank van Bortel wrote:
>> EscVector schreef:
>>
>>> "Do readers block writers in sqlserver".
>>>
>>> I can get this to happen in Oracle as well.
>>> Very easy to do do in fact. All you need is parallel query/partitioned
>>> table and toad or any other sql query tool. Child process remain
>>> active after simple select and block additional queries who's sessions
>>> need that particular cpu. I can reproduce this in 9i every time.
>> Show it - it's not blocking; maybe it's busy.... Not blocking
>> (and in fact what you describe is readers blocking readers)
>> --
>> Regards,
>> Frank van Bortel
>>
>> Top-posting is one way to shut me up...
>
> " I can reproduce this in 9i every time."
> I resend this statement. Was a little pre-new-year over zealousness.
>
> This was on 9.2.0.4. I can't replicate in my current environment
> yet.....
>
> Statement should be "I could reproduce this on 9.2.0.4 with particular
> security patches applied, in a particular environment running
> particular jobs, each time on different machines running different os
> versions, thus making it appear that a "reader" was blocking writers or
> anything that required a particular cpu that had been used by a slave
> process until parent process moved on to another action."
>
> Also, I can't recreate it on 10.2.0.2 with dell winxp laptop with
> simple parallel select as I had done in 9.2.0.4 at previous client.
> Could be combo insert/delete + parallel that caused the problem.
>
> but it did happen, I promise, it really did..... A singularity perhaps?
>
Parallel? As in parallel universe ;)
Happy 2007!
--
Regards,
Frank van Bortel
Top-posting is one way to shut me up...
Received on Sat Dec 30 2006 - 01:30:14 CST
