RE: Skipping logical replication transactions on subscriber side
Zhijie Hou (Fujitsu) <houzj.fnst@fujitsu.com>
On Tuesday, August 3, 2021 2:49 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
>
> I've attached new patches that incorporate all comments I got so far.
> Please review them.
Hi,
I had a few comments for the 0003 patch.
1).
- This clause alters parameters originally set by
- <xref linkend="sql-createsubscription"/>. See there for more
- information. The parameters that can be altered
- are <literal>slot_name</literal>,
- <literal>synchronous_commit</literal>,
- <literal>binary</literal>, and
- <literal>streaming</literal>.
+ This clause sets or resets a subscription option. The parameters that can be
+ set are the parameters originally set by <xref linkend="sql-createsubscription"/>:
+ <literal>slot_name</literal>, <literal>synchronous_commit</literal>,
+ <literal>binary</literal>, <literal>streaming</literal>.
+ </para>
+ <para>
+ The parameters that can be reset are: <literal>streaming</literal>,
+ <literal>binary</literal>, <literal>synchronous_commit</literal>.
Maybe the doc looks better like the following ?
+ This clause alters parameters originally set by
+ <xref linkend="sql-createsubscription"/>. See there for more
+ information. The parameters that can be set
+ are <literal>slot_name</literal>,
+ <literal>synchronous_commit</literal>,
+ <literal>binary</literal>, and
+ <literal>streaming</literal>.
+ </para>
+ <para>
+ The parameters that can be reset are: <literal>streaming</literal>,
+ <literal>binary</literal>, <literal>synchronous_commit</literal>.
2).
- opts->create_slot = defGetBoolean(defel);
+ if (!is_reset)
+ opts->create_slot = defGetBoolean(defel);
}
Since we only support RESET streaming/binary/synchronous_commit, it
might be unnecessary to add the check 'if (!is_reset)' for other
option.
3).
typedef struct AlterSubscriptionStmt
{
NodeTag type;
AlterSubscriptionType kind; /* ALTER_SUBSCRIPTION_OPTIONS, etc */
Since the patch change the remove the enum value
'ALTER_SUBSCRIPTION_OPTIONS', it'd better to change the comment here
as well.
Best regards,
houzj
Commits
-
Test ALIGNOF_DOUBLE==4 compatibility under ALIGNOF_DOUBLE==8.
- c1da0acbb06e 15.0 landed
-
Reorder subskiplsn in pg_subscription to avoid alignment issues.
- 79b716cfb7a1 15.0 landed
-
Add ALTER SUBSCRIPTION ... SKIP.
- 208c5d65bbd6 15.0 landed
-
Optionally disable subscriptions on error.
- 705e20f8550c 15.0 cited
-
Update docs of logical replication for commit 8d74fc96db.
- 85c61ba8920b 15.0 landed
-
Respect permissions within logical replication.
- a2ab9c06ea15 15.0 cited
-
Fix regression test failure caused by commit 8d74fc96db.
- 41e66fee0516 15.0 landed
-
Add a view to show the stats of subscription workers.
- 8d74fc96db5f 15.0 landed
-
Add logical change details to logical replication worker errcontext.
- abc0910e2e0a 15.0 landed
-
Rename LOGICAL_REP_MSG_STREAM_END to LOGICAL_REP_MSG_STREAM_STOP.
- 4cd7a1896871 15.0 landed
-
Fix typo in protocol.sgml.
- e1915646658d 14.0 landed
- 0ac1aee0d7d8 15.0 landed
-
Remove unused argument in apply_handle_commit_internal().
- f4b939f1a372 14.0 landed
- 16bd4becee32 15.0 landed
-
Fix replication of in-progress transactions in tablesync worker.
- 0926e96c4934 14.0 cited
-
Reorder pg_sequence columns to avoid alignment issue
- f3b421da5f4a 10.0 cited