Horton

Re: [PoC] Improve dead tuple storage for lazy vacuum

John Naylor <johncnaylorls@gmail.com>

View thread View on postgr.es
From: John Naylor <johncnaylorls@gmail.com>
To: Masahiko Sawada <sawada.mshk@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>, Nathan Bossart <nathandbossart@gmail.com>, Andres Freund <andres@anarazel.de>, Matthias van de Meent <boekewurm+postgres@gmail.com>, Yura Sokolov <y.sokolov@postgrespro.ru>
Date: 2024-03-28T09:15:28Z
Lists: pgsql-hackers 
On Thu, Mar 28, 2024 at 12:55 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
>
> Pushed the refactoring patch.
>
> I've attached the rebased vacuum improvement patch for cfbot. I
> mentioned in the commit message that this patch eliminates the 1GB
> limitation.
>
> I think the patch is in good shape. Do you have other comments or
> suggestions, John?

I'll do another pass tomorrow, but first I wanted to get in another
slightly-challenging in-situ test. On my humble laptop, I can still
fit a table large enough to cause PG16 to choke on multiple rounds of
index cleanup:

drop table if exists test;
create unlogged table test (a int, b uuid) with (autovacuum_enabled=false);

insert into test (a,b) select i, gen_random_uuid() from
generate_series(1,1000*1000*1000) i;

create index on test (a);
create index on test (b);

delete from test;

vacuum (verbose, truncate off, parallel 2) test;

INFO:  vacuuming "john.public.test"
INFO:  launched 1 parallel vacuum worker for index vacuuming (planned: 1)
INFO:  finished vacuuming "john.public.test": index scans: 1
pages: 0 removed, 6369427 remain, 6369427 scanned (100.00% of total)
tuples: 999997174 removed, 2826 remain, 0 are dead but not yet removable
tuples missed: 2826 dead from 18 pages not removed due to cleanup lock
contention
removable cutoff: 771, which was 0 XIDs old when operation ended
new relfrozenxid: 767, which is 4 XIDs ahead of previous value
frozen: 0 pages from table (0.00% of total) had 0 tuples frozen
index scan needed: 6369409 pages from table (100.00% of total) had
999997174 dead item identifiers removed
index "test_a_idx": pages: 2741898 in total, 2741825 newly deleted,
2741825 currently deleted, 0 reusable
index "test_b_idx": pages: 3850387 in total, 3842056 newly deleted,
3842056 currently deleted, 0 reusable
avg read rate: 159.740 MB/s, avg write rate: 161.726 MB/s
buffer usage: 26367981 hits, 14958634 misses, 15144601 dirtied
WAL usage: 3 records, 1 full page images, 2050 bytes
system usage: CPU: user: 151.89 s, system: 193.54 s, elapsed: 731.59 s

Watching pg_stat_progress_vacuum, dead_tuple_bytes got up to 398458880.

About the "tuples missed" -- I didn't expect contention during this
test. I believe that's completely unrelated behavior, but wanted to
mention it anyway, since I found it confusing.

Commits

  1. radixtree: Fix SIGSEGV at update of embeddable value to non-embeddable.

  2. Get rid of anonymous struct

  3. Teach radix tree to embed values at runtime

  4. Teach TID store to skip bitmap for small numbers of offsets

  5. Use bump context for TID bitmaps stored by vacuum

  6. Fix alignment of stack variable

  7. Use TidStore for dead tuple TIDs storage during lazy vacuum.

  8. Rethink create and attach APIs of shared TidStore.

  9. Fix inconsistent function prototypes with function definitions.

  10. Fix a calculation in TidStoreCreate().

  11. Fix potential integer handling issue in radixtree.h.

  12. Add TIDStore, to store sets of TIDs (ItemPointerData) efficiently.

  13. Fix link error for test_radixtree module on Windows

  14. Blind attempt to fix ODR violations

  15. Fix incorrect format specifier for int64

  16. Fix redefinition of typedefs

  17. Add template for adaptive radix tree

  18. Fix signedness error in 9f225e992 for gcc

  19. Introduce helper SIMD functions for small byte arrays

  20. Optimize vacuuming of relations with no indexes.

  21. Add bound check before bsearch() for performance

  22. Allocate consecutive blocks during parallel seqscans