Tarantool development patches archive
 help / color / mirror / Atom feed
From: Vladimir Davydov <vdavydov.dev@gmail.com>
To: Georgy Kirichenko <georgy@tarantool.org>
Cc: tarantool-patches@freelists.org
Subject: Re: [tarantool-patches] [PATCH v3 09/14] wal: a dedicated wal scheduling fiber
Date: Thu, 13 Jun 2019 17:24:52 +0300	[thread overview]
Message-ID: <20190613142452.j63yqr4q53xszo4r@esperanza> (raw)
In-Reply-To: <64f63cc1cea7a3c632561f36eb043cb386c4e472.1560112747.git.georgy@tarantool.org>

On Sun, Jun 09, 2019 at 11:44:38PM +0300, Georgy Kirichenko wrote:
> In order to implement asynchronous transaction we need to run a
> transaction completion handler but tx_prio is not able to yield.
> 
> Prerequisites: #1254
> ---
>  src/box/wal.c | 48 +++++++++++++++++++++++++++++++++++++++++-------
>  1 file changed, 41 insertions(+), 7 deletions(-)
> 
> diff --git a/src/box/wal.c b/src/box/wal.c
> index 86d021896..e868a8e71 100644
> --- a/src/box/wal.c
> +++ b/src/box/wal.c
> @@ -100,6 +100,12 @@ struct wal_writer
>  	struct cpipe wal_pipe;
>  	/** A memory pool for messages. */
>  	struct mempool msg_pool;
> +	/** A queue to schedule journal entry completions. */
> +	struct stailq schedule_queue;
> +	/** True if writer is in rollback state. */
> +	bool is_in_rollback;
> +	/** A condition to signal about new schedule queue entries. */
> +	struct fiber_cond schedule_cond;
>  	/* ----------------- wal ------------------- */
>  	/** A setting from instance configuration - rows_per_wal */
>  	int64_t wal_max_rows;
> @@ -254,17 +260,36 @@ xlog_write_entry(struct xlog *l, struct journal_entry *entry)
>  	return xlog_tx_commit(l);
>  }
>  
> +/*
> + * Tx schedule fiber function.
> + */
> +static int
> +tx_schedule_f(va_list ap)
> +{
> +	struct wal_writer *writer = va_arg(ap, struct wal_writer *);
> +	while (!fiber_is_cancelled()) {
> +		while (!stailq_empty(&writer->schedule_queue)) {
> +			struct journal_entry *req =
> +				stailq_shift_entry(&writer->schedule_queue,
> +						   struct journal_entry, fifo);
> +			req->done = true;
> +			fiber_cond_broadcast(&req->done_cond);
> +		}
> +		writer->is_in_rollback = false;
> +		fiber_cond_wait(&writer->schedule_cond);
> +	}
> +	return 0;
> +}
> +
>  /**
> - * Signal done condition.
> + * Attach requests to a scheduling queue.
>   */
>  static void
>  tx_schedule_queue(struct stailq *queue)
>  {
> -	struct journal_entry *req;
> -	stailq_foreach_entry(req, queue, fifo) {
> -		req->done = true;
> -		fiber_cond_broadcast(&req->done_cond);
> -	}
> +	struct wal_writer *writer = &wal_writer_singleton;
> +	stailq_concat(&writer->schedule_queue, queue);
> +	fiber_cond_signal(&writer->schedule_cond);

This adds an extra ctxsw to the relatively hot WAL writer wakeup path so
we can't commit it. I assume it's a temporary hack you added so that you
can submit this patches for review.

I understand why you're doing this, but the right way would be removing
yields from on_commit triggers. We need to make vy_log_write yield-free.

  reply	other threads:[~2019-06-13 14:24 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-09 20:44 [tarantool-patches] [PATCH v3 00/14] Parallel applier Georgy Kirichenko
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 01/14] txn: Fire a trigger after a transaction finalization Georgy Kirichenko
2019-06-09 21:59   ` [tarantool-patches] " Konstantin Osipov
2019-06-11 11:42   ` [tarantool-patches] " Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 02/14] ddl: synchronize privileges cache with actual data state Georgy Kirichenko
2019-06-11 13:13   ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 03/14] txn: transaction memory allocation Georgy Kirichenko
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 04/14] ddl: place alter structures onto a txn memory region Georgy Kirichenko
2019-06-11 14:14   ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 05/14] txn: get rid of autocommit from a txn structure Georgy Kirichenko
2019-06-13 14:11   ` Vladimir Davydov
2019-06-16 16:20     ` [tarantool-patches] " Konstantin Osipov
2019-06-16 16:14   ` Konstantin Osipov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 06/14] txn: get rid of fiber_gc from txn_rollback Georgy Kirichenko
2019-06-13 14:12   ` Vladimir Davydov
2019-06-13 19:28     ` Георгий Кириченко
2019-06-14  9:21       ` Vladimir Davydov
2019-06-16 16:38   ` [tarantool-patches] " Konstantin Osipov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 07/14] wal: remove fiber from a journal_entry structure Georgy Kirichenko
2019-06-13 14:17   ` Vladimir Davydov
2019-06-13 19:33     ` Георгий Кириченко
2019-06-14  8:05       ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 08/14] wal: enable asyncronous wal writes Georgy Kirichenko
2019-06-13 14:21   ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 09/14] wal: a dedicated wal scheduling fiber Georgy Kirichenko
2019-06-13 14:24   ` Vladimir Davydov [this message]
2019-06-13 19:36     ` Георгий Кириченко
2019-06-14  9:20       ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 10/14] core: latch_unlock_external routine Georgy Kirichenko
2019-06-13 14:27   ` Vladimir Davydov
2019-06-13 19:38     ` Георгий Кириченко
2019-06-14  8:10       ` Vladimir Davydov
2019-06-14  9:18         ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 11/14] txn: introduce asynchronous txn commit Georgy Kirichenko
2019-06-13 14:34   ` Vladimir Davydov
2019-06-13 19:45     ` Георгий Кириченко
2019-06-14  7:58       ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 12/14] txn: handle fiber stop event at transaction level Georgy Kirichenko
2019-06-13 14:36   ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 13/14] applier: apply transaction in parallel Georgy Kirichenko
2019-06-13 15:17   ` Vladimir Davydov
2019-06-09 20:44 ` [tarantool-patches] [PATCH v3 14/14] test: fix flaky test Georgy Kirichenko

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190613142452.j63yqr4q53xszo4r@esperanza \
    --to=vdavydov.dev@gmail.com \
    --cc=georgy@tarantool.org \
    --cc=tarantool-patches@freelists.org \
    --subject='Re: [tarantool-patches] [PATCH v3 09/14] wal: a dedicated wal scheduling fiber' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox