On 2024/07/11 1:35, Robert Haas wrote:
> On Wed, Jul 10, 2024 at 10:10 AM Robert Haas <robertmhaas@gmail.com> wrote:
>> On Wed, Jul 10, 2024 at 1:56 AM Fujii Masao <masao.fujii@oss.nttdata.com> wrote:
>>> I believe this issue occurs when the server is shut down cleanly.
>>> The shutdown-checkpoint record retains the wal_level value used
>>> before the shutdown. If wal_level is changed after this,
>>> the wal_level that indicated by the shutdown-checkpoint record
>>> and that the WAL data generated afterwards depends on may not match.
>>
>> Oh, that's a problem. I'll have to think about that.
>
> Here is an attempt at fixing this problem.
Thanks for updating the patch!
+ * fast_forward is normally false, but is true when we have encountered
+ * WAL generated with wal_level=minimal and are skipping over it without
+ * emitting summary files. In this case, summarized_tli and summarized_lsn
+ * will advance even though nothing is being written to disk, until we
+ * again reach a point where wal_level>minimal.
+ *
* summarizer_pgprocno is the proc number of the summarizer process, if
* one is running, or else INVALID_PROC_NUMBER.
*
@@ -83,6 +89,7 @@ typedef struct
TimeLineID summarized_tli;
XLogRecPtr summarized_lsn;
bool lsn_is_exact;
+ bool fast_forward;
It looks like the fast_forward field in WalSummarizerData is no longer necessary.
So far, I haven't found any other issues with the patch.
Regards,
--
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION