Hi,
I pushed this version! Thanks all, for the contribution and reviews.
> > I would either put WalUsage into PgStat_PendingWalStats (so that it has
> > all the same members as PgStat_WalStats), or figure out a way to
> > maintain WalUsage separately from PgStat_WalStats or something else.
> > W
Hi,
Thanks for the review.
On Fri, 17 Mar 2023 at 02:02, Melanie Plageman
wrote:
> I think you want one less L here?
> WALLSTAT_ACC_INSTR_TIME_TYPE -> WALSTAT_ACC_INSTR_TIME_TYPE
Done.
> Also, I don't quite understand why TYPE is at the end of the name. I
> think it would still be clear withou
On Thu, Mar 09, 2023 at 04:02:44PM +0300, Nazir Bilal Yavuz wrote:
> From dcd49e48a0784a95b8731df1c6ee7c3a612a8529 Mon Sep 17 00:00:00 2001
> From: Nazir Bilal Yavuz
> Date: Thu, 9 Mar 2023 15:35:38 +0300
> Subject: [PATCH v5] Refactor instr_time calculations
>
> Also, some calculations are refac
Hi,
There was a warning while applying the patch, v5 is attached.
Regards,
Nazir Bilal Yavuz
Microsoft
From dcd49e48a0784a95b8731df1c6ee7c3a612a8529 Mon Sep 17 00:00:00 2001
From: Nazir Bilal Yavuz
Date: Thu, 9 Mar 2023 15:35:38 +0300
Subject: [PATCH v5] Refactor instr_time calculations
In inst
Hi,
Thanks for the review.
On Wed, 22 Feb 2023 at 05:50, Kyotaro Horiguchi wrote:
>
> PgStat_PendingStats should be included in typedefs.list.
Done.
>
> + * Created for accumulating wal_write_time and wal_sync_time as a instr_time
> + * but instr_time can't be used as a type where it ends up o
At Tue, 21 Feb 2023 16:11:19 +0300, Nazir Bilal Yavuz
wrote in
> I agree. The patch is updated.
Thanks, that part looks good to me. I'd like to provide some
additional comments.
PgStat_PendingStats should be included in typedefs.list.
+ * Created for accumulating wal_write_time and wal_sync_
Hi,
Thanks for the review.
On Mon, 20 Feb 2023 at 06:01, Kyotaro Horiguchi wrote:
>
> At Fri, 17 Feb 2023 13:53:36 +0300, Nazir Bilal Yavuz
> wrote in
> > Thanks for the review. I updated the patch.
>
>
> WalUsageAccumDiff(&diff, &pgWalUsage, &prevWalUsage);
> - PendingWalStats.w
At Fri, 17 Feb 2023 13:53:36 +0300, Nazir Bilal Yavuz
wrote in
> Thanks for the review. I updated the patch.
WalUsageAccumDiff(&diff, &pgWalUsage, &prevWalUsage);
- PendingWalStats.wal_records = diff.wal_records;
- PendingWalStats.wal_fpi = diff.wal_fpi;
- PendingWalS
Hi,
On 2/16/23 19:13, Andres Freund wrote:
+#define WALSTAT_ACC(fld, var_to_add) \
+ (stats_shmem->stats.fld += var_to_add.fld)
+#define WALLSTAT_ACC_INSTR_TIME_TYPE(fld) \
+ (stats_shmem->stats.fld += INSTR_TIME_GET_MICROSEC(PendingWalStats.fld))
+ WALSTAT_ACC(wal_records, di
Hi,
On 2023-02-16 16:19:02 +0300, Nazir Bilal Yavuz wrote:
> What do you think?
Here's a small review:
> +#define WALSTAT_ACC(fld, var_to_add) \
> + (stats_shmem->stats.fld += var_to_add.fld)
> +#define WALLSTAT_ACC_INSTR_TIME_TYPE(fld) \
> + (stats_shmem->stats.fld += INSTR_TIME_GET_MIC
Hi,
In 'instr_time.h' it is stated that:
* When summing multiple measurements, it's recommended to leave the
* running sum in instr_time form (ie, use INSTR_TIME_ADD or
* INSTR_TIME_ACCUM_DIFF) and convert to a result format only at the end.
So, I refactored 'PendingWalStats' to use 'instr_ti
11 matches
Mail list logo
