[PATCH v3 0/5] Delete old WAL files if running out of disk space
vdavydov.dev at gmail.com
Thu Oct 25 15:55:19 MSK 2018
On Wed, Oct 24, 2018 at 04:43:12PM +0300, Vladimir Davydov wrote:
> Vladimir Davydov (5):
> wal: preallocate disk space before writing rows
> wal: pass wal_watcher_msg to wal_watcher callback
> wal: rename wal_watcher->events to pending_events
> wal: add event_mask to wal_watcher
> wal: delete old wal files when running out of disk space
Discussed verbally with Kostja. Agreed that even though there are some
concerns regarding the watcher API re-usability and efficiency, we shall
push it as is now and rework it later if we need to.
I added a comment to patch #4 (see below) and pushed all the patches to
+ * Notify all interested watchers about a WAL event.
+ * XXX: Note, this function iterates over all registered watchers,
+ * including those that are not interested in the given event.
+ * This is OK only as long as the number of events/watchers is
+ * small. If this ever changes, we should consider maintaining
+ * a separate watcher list per each event type.
wal_notify_watchers(struct wal_writer *writer, unsigned events)
More information about the Tarantool-patches