Re: Two fsync related performance issues?

2020-05-11 Thread Michael Paquier
On Tue, May 12, 2020 at 12:55:37PM +0900, Fujii Masao wrote: > On 2020/05/12 9:42, Paul Guo wrote: >> 1. StartupXLOG() does fsync on the whole data directory early in >> the crash recovery. I'm wondering if we could skip some >> directories (at least the pg_log/, table directories) since wal, >> et

Re: A comment fix

2020-05-11 Thread Kyotaro Horiguchi
At Tue, 12 May 2020 14:45:22 +0900, Michael Paquier wrote in > On Mon, May 11, 2020 at 02:22:36PM +0900, Michael Paquier wrote: > > Looks right to me, so will fix if there are no objections. > > read_local_xlog_page() uses the replay location when in recovery. > > Done this part now. Thanks!

Re: tablespace_map code cleanup

2020-05-11 Thread Amit Kapila
On Thu, May 7, 2020 at 9:44 PM Robert Haas wrote: > > On Wed, May 6, 2020 at 11:15 AM Robert Haas wrote: > > Oh, good point. v2 attached. > While looking at this, I noticed that caller (perform_base_backup) of do_pg_start_backup, sets the backup phase as PROGRESS_BASEBACKUP_PHASE_WAIT_CHECKPOINT

Re: Event trigger code comment duplication

2020-05-11 Thread Michael Paquier
On Mon, May 11, 2020 at 05:13:38PM -0700, David G. Johnston wrote: > Skimming through the code in event_trigger.c and noticed that while most of > the stanzas that reference IsUnderPostmaster refer back to the code comment > beginning on line 675 the block for table rewrite copied it in > verbatim

<    1   2