On Tue, Nov 8, 2016 at 5:18 PM, Kyotaro HORIGUCHI
<horiguchi.kyotaro@lab.ntt.co.jp> wrote:
> Hello,
>
>> on something else than LW_EXCLUSIVE compared to now. To keep things
>> more simple I' would still favor using WALInsertLocks for this patch,
>> that looks more consistent, and also because there is no noticeable
>> difference.
>
> Ok, the patch looks fine. So there's nothing for me but to accept
> the current shape since the discussion about performance seems
> not to be settled with out performance measurement with machines
> with many cores.
>
I think it is good to check the performance impact of this patch on
many core m/c. Is it possible for you to once check with Alexander
Korotkov to see if he can provide you access to his powerful m/c which
has 70 cores (if I remember correctly)?
@@ -1035,6 +1038,7 @@ LogAccessExclusiveLocks(int nlocks,
xl_standby_lock *locks) XLogBeginInsert(); XLogRegisterData((char *) &xlrec, offsetof(xl_standby_locks, locks));
XLogRegisterData((char*) locks, nlocks * sizeof(xl_standby_lock));
+ XLogSetFlags(XLOG_NO_PROGRESS);
Is it right to set XLOG_NO_PROGRESS flag in LogAccessExclusiveLocks?
This function is called not only in LogStandbySnapshot(), but during
DDL operations as well where lockmode >= AccessExclusiveLock.
--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com