SEO

#1 (**permalink**) 04-10-2008, 05:04 AM

Hi,

I managed to reproduce a locking problem, which is about blocking
transactions and dead connections. How can this be avoided?

What I did was:
1. Start a transaction on Client A
2. Execute a fcn doing an UPDATE on a certain record on client A
3. Execute a fcn doing an UPDATE on the same record on client B
/ as expected SQL on client B stalls untl client A's transaction ends/
4. disconnect client A's network cable
5. kill the psql on client A
/ C.A's transaction still runs since two hours, C.B's SQL lasts for equal
that time/
6. wait on client B's statement to finish...
7. kill -TERM client A's postmaster
/ client B's statement succeeds

Why is that? Why isn't client A's transaction detected as dead ? Yes,
because the socked wasn't closed due to a network error. But this can happen
all the time. What can I do?

---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faq

#2 (**permalink**) 04-10-2008, 05:04 AM

=?iso-8859-1?Q?K=D6PFERL_Robert?= <robert.koepferl@sonorys.at> writes:
> Why is that? Why isn't client A's transaction detected as dead ?

It will be eventually, when the TCP connection times out. The standard
timeout is generally an hour or two :-(

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

SEO

Blocking connection and timeout problem