[pgpool-general: 4277] Re: Pgpool - connection hangs in DISCARD ALL

[Gerhard Wiesinger]

On 31.12.2015 08:48, Tatsuo Ishii wrote:
>>>> It worked now for 3 days well but today I still had the same
>>>> issue. Looks like the probability has been reduced, but not fixed in
>>>> all cases.
>>> [snip]
>>>> root 1349 0.0 0.1 94248 864 ?  S 00:13 0:01 pgpool: nag hei
>>>> 192.168.0.9(39886) DISCARD
>>> Yeah, looks like pgpool is hanging while executing DISCARD. Assuming
>>> you apply the fix in 3.4 stable tree, do you know how to reproduce the
>>> problem?
>> Yes, patch is applied. I verified it again on the RPM build directory.
>>
>> No, I don't know how to reproduce. It just happens on a production
>> machine. But I attached a detailed stack trace in the previous post.
> I suspect the logic for child_max_connections is not good. Can you
> disable child_max_connections (i.e. child_max_connections = 0)?

Yes, reconfigured. Keep you up2date. Any fix for the logic possible?

BTW: Maybe this has already been discussed but if I understood it 
correctly there is no global pool available.
http://www.pgpool.net/mediawiki/index.php/FAQ#Why_max_connection_must_satisfy_this_formula_max_connection_.3E.3D_.28num_init_children_.2A_max_pool.29_and_not_max_connection_.3E.3D_num_init_children.3F
That means if the wrong client is selected a new backend connection is 
opened. Even when another idle client would have another idle backend 
already opened.
Wouldn't it be better to have a global backend pool?

Also ressource usage is much higher (num_init_children * max_pool).

Any ideas on that or possible changes?

Ciao,
Gerhard