Thread: transaction-lock?
Hi,
I am a novice pgsql user. I used Oracle, but it is unbelievable what
bugs they have...
So, is it possible, that if one transaction is pending on for eg. 'test'
table in one psql session, then in another session this 'test' table is
locked until the transaction is not finished on another????
I have been tested a lot of ways, but the most easier problem is the
next:
- there is a table named 'test', with only 1 field.
- opening a psql session on tty1, and on tty2 into the same database
- on tty1 'begin' a transaction
- on tty1 'insert' 1 row into the 'test' table
- on tty2 'select' from the 'test' table
=> on the tty2 there is no answer until on the tty1 is the
transaction not finished
This runs on the next architecture (compiled today the new 6.4
release!):
- IP133, 49M RAM
- Debian 2.0
If I have been a inattentive then please send me the solution, and
sorry; if not, then please send me the solution, and 'not at all'.
FT
At 20:06 +0200 on 23/03/1999, Forgacs Tamas wrote: > > I am a novice pgsql user. I used Oracle, but it is unbelievable what > bugs they have... > > So, is it possible, that if one transaction is pending on for eg. 'test' > table in one psql session, then in another session this 'test' table is > locked until the transaction is not finished on another???? > > I have been tested a lot of ways, but the most easier problem is the > next: > - there is a table named 'test', with only 1 field. > - opening a psql session on tty1, and on tty2 into the same database > > - on tty1 'begin' a transaction > - on tty1 'insert' 1 row into the 'test' table > - on tty2 'select' from the 'test' table > => on the tty2 there is no answer until on the tty1 is the > transaction not finished Yes, that's the documented behaviour. Oracle has bugs, but Postgres will have a better locking resolution only starting 6.5. At the moment it locks on tables. Herouth -- Herouth Maoz, Internet developer. Open University of Israel - Telem project http://telem.openu.ac.il/~herutma