RE: Rework LogicalOutputPluginWriterUpdateProgress - Mailing list pgsql-hackers

From Takamichi Osumi (Fujitsu)
Subject RE: Rework LogicalOutputPluginWriterUpdateProgress
Date
Msg-id TYCPR01MB83731C1B9C85220228498F13EDBA9@TYCPR01MB8373.jpnprd01.prod.outlook.com
Whole thread Raw
In response to RE: Rework LogicalOutputPluginWriterUpdateProgress  ("wangw.fnst@fujitsu.com" <wangw.fnst@fujitsu.com>)
Responses RE: Rework LogicalOutputPluginWriterUpdateProgress  ("wangw.fnst@fujitsu.com" <wangw.fnst@fujitsu.com>)
List pgsql-hackers
Hi,


On Friday, March 10, 2023 6:32 PM Wang, Wei/王 威 <wangw.fnst@fujitsu.com> wrote:
> Attach the new patch set.
Thanks for updating the patch ! One review comment on v7-0005.

stream_start_cb_wrapper and stream_stop_cb_wrapper don't call the pair of threshold check and
UpdateProgressAndKeepaliveunlike other write wrapper functions like below. But, both of them write some data to the
outputplugin, set the flag of did_write and thus it updates the subscriber's last_recv_timestamp used for timeout check
inLogicalRepApplyLoop. So, it looks adding the pair to both functions can be more accurate, in order to reset the
counterin changes_count on the publisher ?
 

@@ -1280,6 +1282,8 @@ stream_start_cb_wrapper(ReorderBuffer *cache, ReorderBufferTXN *txn,

        /* Pop the error context stack */
        error_context_stack = errcallback.previous;
+
+       /* No progress has been made, so don't call UpdateProgressAndKeepalive */
 }


Best Regards,
    Takamichi Osumi


pgsql-hackers by date:

Previous
From: Amit Kapila
Date:
Subject: Re: [PATCH] Use indexes on the subscriber when REPLICA IDENTITY is full on the publisher
Next
From: Pavel Luzanov
Date:
Subject: Re: psql: Add role's membership options to the \du+ command