At Tue, 3 Aug 2021 20:46:57 +0000, "Bossart, Nathan" <bossa...@amazon.com> 
wrote in 
> +     /*
> +      * Perform a full directory scan to identify the next log segment. There
> +      * may be one of the following scenarios which may require us to 
> perform a
> +      * full directory scan.
> +      *
> +      * 1. This is the first cycle since archiver has started and there is no
> +      * idea about the next anticipated log segment.
> +      *
> +      * 2. There is a timeline switch, i.e. the timeline ID tracked at 
> archiver
> +      * does not match with current timeline ID. Archive history file as 
> part of
> +      * this timeline switch.
> +      *
> +      * 3. The next anticipated log segment is not available.
> 
> One benefit of the current implementation of pgarch_readyXlog() is
> that .ready files created out of order will be prioritized before
> segments with greater LSNs.  IIUC, with this patch, as long as there
> is a "next anticipated" segment available, the archiver won't go back
> and archive segments it missed.  I don't think the archive status
> files are regularly created out of order, but XLogArchiveCheckDone()
> has handling for that case, and the work to avoid creating .ready
> files too early [0] seems to make it more likely.  Perhaps we should
> also force a directory scan when we detect that we are creating a
> .ready file for a segment that is older than the "next anticipated"
> segment.
> 
> Nathan
> 
> [0] https://postgr.es/m/DA71434B-7340-4984-9B91-F085BC47A778%40amazon.com
It works the current way always at the first iteration of
pgarch_ArchiveCopyLoop() becuse in the last iteration of
pgarch_ArchiveCopyLoop(), pgarch_readyXlog() erases the last
anticipated segment.  The shortcut works only when
pgarch_ArchiveCopyLoop archives more than once successive segments at
once.  If the anticipated next segment found to be missing a .ready
file while archiving multiple files, pgarch_readyXLog falls back to
the regular way.

So I don't see the danger to happen perhaps you are considering.

In the first place, .ready are added while holding WALWriteLock in
XLogWrite, and while removing old segments after a checkpoint (which
happens while recovery). Assuming that no one manually remove .ready
files on an active server, the former is the sole place doing that. So
I don't see a chance that .ready files are created out-of-order way.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center


Reply via email to