On Thu, 18 Sep 2003, Ravi T Ramachandra wrote:
>
> Hi,
>
> We are using postgresql 7.3.2 on lunux platform. We performed locking
> test as follows via JDBC:
>
> Started 2 concurrent transaction with autocommit false and isolation
> READ_COMMITTED :
>
> Transaction 1:
> ===========
> Update row 1
> Sleep for 30 seconds
> commit:
>
> Transaction 2:
> ===========
> Updated row 10000
> commit
>
> The second update had to wait for the first update operation to finish.
> We also made sure that the 2 unrelated rows are not on the same page.
>
> Is this expected behavior ?
> Does update lock the whole table ?
Generally not, but you really haven't given alot of information. What is
the table layout for the table in question? Are there foreign keys
involved?