[PATCH 1/2] memtx: add yields during space format check

Vladimir Davydov vdavydov.dev at gmail.com
Tue Jun 4 18:17:45 MSK 2019


On Mon, Jun 03, 2019 at 06:51:05PM +0300, Serge Petrenko wrote:
> Just like index build, space check in memtx stalls event loop for the
> whole check time. Add occasional yields, and an on_replace trigger,
> which checks format for tuples inserted while space format check is in
> progress.
> 
> Follow-up #3976
> 
> @TarantoolBol document
> Title: memtx now checks space format in background
> 
> There is no event loop stall when memtx engine checks space format
> anymore. You may insert tuples into a space, while its new format is
> being checked. If the tuples don't match the new format, format change
> will be aborted.
> ---
>  src/box/memtx_space.c | 110 ++++++++++++++++++++++++++++++++----------
>  1 file changed, 84 insertions(+), 26 deletions(-)
> 
> diff --git a/src/box/memtx_space.c b/src/box/memtx_space.c
> index 1d209033c..a370c038a 100644
> --- a/src/box/memtx_space.c
> +++ b/src/box/memtx_space.c
> @@ -43,6 +43,16 @@
>  #include "column_mask.h"
>  #include "sequence.h"
>  
> +/*
> + * Yield every 1K tuples.
> + * In debug mode yield more often for testing purposes.
> + */
> +#ifdef NDEBUG
> +enum { MEMTX_YIELD_LOOPS = 1000 };
> +#else
> +enum { MEMTX_YIELD_LOOPS = 10 };
> +#endif
> +
>  static void
>  memtx_space_destroy(struct space *space)
>  {
> @@ -814,6 +824,52 @@ memtx_space_add_primary_key(struct space *space)
>  	return 0;
>  }
>  
> +/*
> + * Ongoing index build or format check state used by
> + * corrseponding on_replace triggers.
> + */
> +struct memtx_op_state {

What's 'op'? The name's too generic. I renamed it to memtx_ddl_state.
I also renamed MEMTX_YIELD_LOOPS to MEMTX_DDL_YIELD_LOOPS to emphasize
that this is about DDL.

I pushed both patches to master with this minor changes. Thanks!



More information about the Tarantool-patches mailing list