Re: lock - feedback

Richard Huxton wrote:
> Thomas Rokohl wrote:
>> Richard Huxton wrote:
>>> Thomas Rokohl wrote:
>>>> hi,
>>>>
>>>> i have a short question and i don't find an answer is the archive.
>>>> i'm using the odbc interface to the db and if i lock rows with
>>>> SELECT FOR UPDATE  and someone else try the same or
>>>> try to read the rows, i will give him a feedback that he must wait
>>>> or abort.
>>>> The Problem is that the SELECT statement wait without feedback until
>>>> the end of the world if the row is lock.
>>>
>>> Actually, it will timeout, and you can set that with PG's
>>> configuration setting of "statement_timeout". See the "run-time
>>> configuration" section of the manual for details.
>>>
>> first: thanks for your fast answer.
>>
>> if had found it in the manual but there is only a short clause about
>> this topic:
>>
>> "Abort any statement that takes over the specified number of
>> milliseconds. A value of zero (the default) turns off the limitation. "
>>
>> ( by the way in my config file is it set to zero by default and that
>> means that it will be wait without a timeout.
>> #statement_timeout = 0        # 0 is disabled, in milliseconds
>> )
>>
>> so but i'm not sure, doesn't it mean that all statements that take
>> longer than this time will be abort?
>> if it is so, that isn't a solution for this problem because than i
>> can't make a "big" request anymore.
>
> You can change it during a session "set statement_timeout=2000"
>
yes, but that isn't a certain event that say that the data is locked!
also the network for example can determine different timeouts.
and also i don't know the working time for each request before.
that isn't a solution for me. i will open a "please wait, while the data
is locked" - window, if the data is locked.

and it seems as if this should be a problem with postgres :-(