From: Andrey Melnikov <temnota.am@gmail.com>
To: linux-btrfs@vger.kernel.org, Andrey Melnikov <temnota.am@gmail.com>
Subject: btrfs with huge numbers of hardlinks is extremely slow
Date: Fri, 26 Nov 2021 00:56:25 +0300 [thread overview]
Message-ID: <CA+PODjrE4V9hL1bXEEghU6NAFgPgfUu4f75FCQn+0vKUaeu1zg@mail.gmail.com> (raw)
Every night a new backup is stored on this fs with 'rsync
--link-dest=$yestoday/ $today/' - 1402700 hardlinks and 23000
directories are created, 50-100 normal files transferred.
Now, FS contains 351 copies of backup data with 486086495 hardlinks
and ANY operations on this FS take significant time. For example -
simple count hardlinks with
"time find . -type f -links +1 | wc -l" take:
real 28567m33.611s
user 31m33.395s
sys 506m28.576s
19 days 20 hours 10 mins with constant reads from storage 2-4Mb/s.
- BTRFS not suitable for this workload?
- using reflinks helps speedup FS operations?
- readed metadata not cached at all? What BTRFS read 19 days from disks???
Hardware: dell r300 with 2 WD Purple 1Tb disk on LSISAS1068E RAID 1
(without cache).
Linux nlxc 5.14-51412-generic #0~lch11 SMP Wed Oct 13 15:57:07 UTC
2021 x86_64 GNU/Linux
btrfs-progs v5.14.1
# btrfs fi show
Label: none uuid: a840a2ca-bf05-4074-8895-60d993cb5bdd
Total devices 1 FS bytes used 474.26GiB
devid 1 size 931.00GiB used 502.23GiB path /dev/sdb1
# btrfs fi df /srv
Data, single: total=367.19GiB, used=343.92GiB
System, single: total=32.00MiB, used=128.00KiB
Metadata, single: total=135.00GiB, used=130.34GiB
GlobalReserve, single: total=512.00MiB, used=0.00B
# btrfs fi us /srv
Overall:
Device size: 931.00GiB
Device allocated: 502.23GiB
Device unallocated: 428.77GiB
Device missing: 0.00B
Used: 474.26GiB
Free (estimated): 452.04GiB (min: 452.04GiB)
Free (statfs, df): 452.04GiB
Data ratio: 1.00
Metadata ratio: 1.00
Global reserve: 512.00MiB (used: 0.00B)
Multiple profiles: no
Data,single: Size:367.19GiB, Used:343.92GiB (93.66%)
/dev/sdb1 367.19GiB
Metadata,single: Size:135.00GiB, Used:130.34GiB (96.55%)
/dev/sdb1 135.00GiB
System,single: Size:32.00MiB, Used:128.00KiB (0.39%)
/dev/sdb1 32.00MiB
Unallocated:
/dev/sdb1 428.77GiB
next reply other threads:[~2021-11-25 21:59 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-11-25 21:56 Andrey Melnikov [this message]
2021-11-26 5:15 ` btrfs with huge numbers of hardlinks is extremely slow Zygo Blaxell
2021-11-26 8:23 ` Forza
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=CA+PODjrE4V9hL1bXEEghU6NAFgPgfUu4f75FCQn+0vKUaeu1zg@mail.gmail.com \
--to=temnota.am@gmail.com \
--cc=linux-btrfs@vger.kernel.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).