Re: Performance degradation in commit ac1d794

On December 25, 2015 7:10:23 PM GMT+01:00, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>=?UTF-8?B?0JLQsNGB0LjQu9GM0LXQsiDQlNC80LjRgtGA0LjQuQ==?=
><d(dot)vasilyev(at)postgrespro(dot)ru> writes:
>> ��� Samples: 1M of event 'cycles', Event count (approx.):
>816922259995, UID:
>> pgpro
>> Overhead Shared Object Symbol
>
>> 69,72% [kernel] [k] _raw_spin_lock_irqsave
>> 1,43% postgres [.] _bt_compare
>> 1,19% postgres [.] LWLockAcquire
>> 0,99% postgres [.] hash_search_with_hash_value
>> 0,61% postgres [.] PinBuffer
>
>Seems like what you've got here is a kernel bug.

I wouldn't go as far as calling it a kernel bug. Were still doing 300k tps. And were triggering the performance degradation by adding another socket (IIRC) to the poll(2) call.

It certainly be interesting to see the expanded tree below the spinlock. I wonder if this is related to directed wakeups.

Andres

---
Please excuse brevity and formatting - I am writing this on my mobile phone.