From: Vladimir Davydov <vdavydov.dev@gmail.com>
To: kostja@tarantool.org
Cc: tarantool-patches@freelists.org
Subject: [PATCH 05/12] vinyl: do not reallocate tuple formats on alter
Date: Sun, 1 Apr 2018 12:05:32 +0300 [thread overview]
Message-ID: <475ebc439ee72ccd55b1158c354a01b2452a6bd5.1522572161.git.vdavydov.dev@gmail.com> (raw)
In-Reply-To: <cover.1522572160.git.vdavydov.dev@gmail.com>
In-Reply-To: <cover.1522572160.git.vdavydov.dev@gmail.com>
We create new formats for all indexes of the new space in
vinyl_space_commit_alter() while we don't actually need to
do this, because the new formats have already been created
by vy_lsm_new() - all we need to do is reuse them somehow.
This patch does the trick: it implements the swap_index()
space virtual method for vinyl so that it swaps tuple formats
between the old and new spaces.
---
src/box/vinyl.c | 58 ++++++++++++++++++++++-----------------------------------
1 file changed, 22 insertions(+), 36 deletions(-)
diff --git a/src/box/vinyl.c b/src/box/vinyl.c
index fc37283a..f3cef52f 100644
--- a/src/box/vinyl.c
+++ b/src/box/vinyl.c
@@ -1024,30 +1024,9 @@ vinyl_space_commit_alter(struct space *old_space, struct space *new_space)
struct tuple_format *new_format = new_space->format;
struct vy_lsm *pk = vy_lsm(new_space->index[0]);
- struct index_def *new_index_def = space_index_def(new_space, 0);
-
assert(pk->pk == NULL);
- /* Update the format with column mask. */
- struct tuple_format *format =
- vy_tuple_format_new_with_colmask(new_format);
- if (format == NULL)
- goto fail;
-
- /* Set possibly changed opts. */
- pk->opts = new_index_def->opts;
pk->check_is_unique = true;
-
- /* Set new formats. */
- tuple_format_unref(pk->disk_format);
- tuple_format_unref(pk->mem_format);
- tuple_format_unref(pk->mem_format_with_colmask);
- pk->disk_format = new_format;
- tuple_format_ref(new_format);
- pk->mem_format_with_colmask = format;
- tuple_format_ref(format);
- pk->mem_format = new_format;
- tuple_format_ref(new_format);
vy_lsm_validate_formats(pk);
key_def_update_optionality(pk->key_def, new_format->min_field_count);
key_def_update_optionality(pk->cmp_def, new_format->min_field_count);
@@ -1057,15 +1036,7 @@ vinyl_space_commit_alter(struct space *old_space, struct space *new_space)
vy_lsm_unref(lsm->pk);
vy_lsm_ref(pk);
lsm->pk = pk;
- new_index_def = space_index_def(new_space, i);
- lsm->opts = new_index_def->opts;
lsm->check_is_unique = lsm->opts.is_unique;
- tuple_format_unref(lsm->mem_format_with_colmask);
- tuple_format_unref(lsm->mem_format);
- lsm->mem_format_with_colmask = pk->mem_format_with_colmask;
- lsm->mem_format = pk->mem_format;
- tuple_format_ref(lsm->mem_format_with_colmask);
- tuple_format_ref(lsm->mem_format);
key_def_update_optionality(lsm->key_def,
new_format->min_field_count);
key_def_update_optionality(lsm->cmp_def,
@@ -1093,12 +1064,27 @@ vinyl_space_commit_alter(struct space *old_space, struct space *new_space)
}
}
}
- return;
-fail:
- /* FIXME: space_vtab::commit_alter() must not fail. */
- diag_log();
- unreachable();
- panic("failed to alter space");
+}
+
+static void
+vinyl_space_swap_index(struct space *old_space, struct space *new_space,
+ uint32_t old_index_id, uint32_t new_index_id)
+{
+ struct vy_lsm *old_lsm = vy_lsm(old_space->index_map[old_index_id]);
+ struct vy_lsm *new_lsm = vy_lsm(new_space->index_map[new_index_id]);
+
+ /*
+ * Swap the two indexes between the two spaces,
+ * but leave tuple formats.
+ */
+ generic_space_swap_index(old_space, new_space,
+ old_index_id, new_index_id);
+
+ SWAP(old_lsm->mem_format, new_lsm->mem_format);
+ SWAP(old_lsm->mem_format_with_colmask,
+ new_lsm->mem_format_with_colmask);
+ SWAP(old_lsm->disk_format, new_lsm->disk_format);
+ SWAP(old_lsm->opts, new_lsm->opts);
}
static int
@@ -3938,7 +3924,7 @@ static const struct space_vtab vinyl_space_vtab = {
/* .drop_primary_key = */ vinyl_space_drop_primary_key,
/* .check_format = */ vinyl_space_check_format,
/* .build_secondary_key = */ vinyl_space_build_secondary_key,
- /* .swap_index = */ generic_space_swap_index,
+ /* .swap_index = */ vinyl_space_swap_index,
/* .prepare_alter = */ vinyl_space_prepare_alter,
/* .commit_alter = */ vinyl_space_commit_alter,
};
--
2.11.0
next prev parent reply other threads:[~2018-04-01 9:05 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-04-01 9:05 [PATCH 00/12] vinyl: allow to extend key def of non-empty index Vladimir Davydov
2018-04-01 9:05 ` [PATCH 01/12] index: add commit_modify virtual method Vladimir Davydov
2018-04-01 9:05 ` [PATCH 02/12] alter: require rebuild of all secondary vinyl indexes if pk changes Vladimir Davydov
2018-04-01 9:05 ` [PATCH 03/12] alter: do not rebuild secondary indexes on compatible " Vladimir Davydov
2018-04-05 12:30 ` Konstantin Osipov
2018-04-01 9:05 ` [PATCH 04/12] space: make space_swap_index method virtual Vladimir Davydov
2018-04-01 9:05 ` Vladimir Davydov [this message]
2018-04-01 11:12 ` [tarantool-patches] [PATCH 05/12] vinyl: do not reallocate tuple formats on alter v.shpilevoy
2018-04-01 11:24 ` Vladimir Davydov
2018-04-01 9:05 ` [PATCH 06/12] vinyl: zap vy_lsm_validate_formats Vladimir Davydov
2018-04-01 9:05 ` [PATCH 07/12] vinyl: zap vy_mem_update_formats Vladimir Davydov
2018-04-01 9:05 ` [PATCH 08/12] vinyl: remove pointless is_nullable initialization for disk_format Vladimir Davydov
2018-04-01 9:05 ` [PATCH 09/12] vinyl: use source tuple format when copying field map Vladimir Davydov
2018-04-01 9:05 ` [PATCH 10/12] vinyl: rename vy_log_record::commit_lsn to create_lsn Vladimir Davydov
2018-04-01 9:05 ` [PATCH 11/12] vinyl: do not write VY_LOG_DUMP_LSM record to snapshot Vladimir Davydov
2018-04-01 9:05 ` [PATCH 12/12] vinyl: allow to modify key definition if it does not require rebuild Vladimir Davydov
2018-04-02 8:46 ` Vladimir Davydov
2018-04-05 14:32 ` Konstantin Osipov
2018-04-05 14:45 ` Konstantin Osipov
2018-04-05 14:48 ` Konstantin Osipov
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=475ebc439ee72ccd55b1158c354a01b2452a6bd5.1522572161.git.vdavydov.dev@gmail.com \
--to=vdavydov.dev@gmail.com \
--cc=kostja@tarantool.org \
--cc=tarantool-patches@freelists.org \
--subject='Re: [PATCH 05/12] vinyl: do not reallocate tuple formats on alter' \
/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