Re: Is the unfair lwlock behavior intended?

On Tue, May 24, 2016 at 4:39 PM, Ants Aasma <ants(dot)aasma(at)eesti(dot)ee> wrote:
> On Tue, May 24, 2016 at 9:03 AM, Tsunakawa, Takayuki
> <tsunakawa(dot)takay(at)jp(dot)fujitsu(dot)com> wrote:
>> I encountered a strange behavior of lightweight lock in PostgreSQL 9.2. That appears to apply to 9.6, too, as far as I examine the code. Could you tell me if the behavior is intended or needs fix?
>>
>> Simply put, the unfair behavior is that waiters for exclusive mode are overtaken by share-mode lockers who arrive later.
>
> 9.5 had significant LWLock scalability improvements. This might
> improve performance enough so that exclusive lockers don't get
> completely starved. It would be helpful if you could test if it's
> still possible to trigger starvation with the new code.

9.5 didn't just increase the scalability; it also whacked the fairness
aspects of this code around.

Author: Andres Freund <andres(at)anarazel(dot)de>
Branch: master Release: REL9_5_BR [7882c3b0b] 2014-12-25 17:24:30 +0100

Convert the PGPROC->lwWaitLink list into a dlist instead of open coding it.

Besides being shorter and much easier to read it changes the logic in
LWLockRelease() to release all shared lockers when waking up any. This
can yield some significant performance improvements - and the fairness
isn't really much worse than before, as we always allowed new shared
lockers to jump the queue.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company