Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 24 Dec 2023 06:22:52 +0200
From:      Sulev-Madis Silber <madis555@hot.ee>
To:        d@delphij.net, Xin Li <delphij@delphij.net>, freebsd-current@freebsd.org
Subject:   Re: Proposal: Disable compression of newsyslog by default
Message-ID:  <E124EE62-15CA-48C7-BED1-BA21B5587EDF@hot.ee>
In-Reply-To: <ace7d170-acd1-4dc8-8016-a08c261d91f0@delphij.net>
References:  <ace7d170-acd1-4dc8-8016-a08c261d91f0@delphij.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 23 December 2023 09:18:23 EET, Xin Li <delphij@delphij=2Enet> wrote:
>Hi,
>
>Inspired by D42961, I propose that we move forward with disabling the com=
pression by default in newsyslog, as implemented in https://reviews=2Efreeb=
sd=2Eorg/D43169
>
>Historically, newsyslog has compressed rotated log files to save disk spa=
ce=2E This approach was valuable in the early days where storage space was =
limited=2E

it's still limited

>However, the landscape has changed significantly=2E  Modern file systems,=
 such as ZFS, now offer native compression capabilities=2E

not everyone uses them

>Additionally, the widespread availability of larger hard drives has dimin=
ished the necessity for additional compression=2E

but data sizes also have increased massively

>Notably, the need to decompress log files for pattern searches poses a si=
gnificant inconvenience, further questioning the utility of this legacy fea=
ture=2E

should be up to each admin to cba decompression vs=2E plain speed/size/etc

>In commit 906748d208d3, flags J, X, Y, Z can now indicate that a log file=
 is eligible for compression rather than directly enforcing it=2E It allows=
 for a more flexible approach, wherein the actual compression method can be=
 set to "none" or specified as one among bzip2, gzip, xz, or zstd=2E

that's good approach

>Therefore I would propose that we change the default compression setting =
to "none" in FreeBSD 15=2E0=2E  This change reflects our adaptation to the =
evolving technological environment and user needs=2E  It also aligns with t=
he broader initiative to modernize our systems while maintaining flexibilit=
y and efficiency=2E

unsure about this=2E generic zroot install would be fine with this i guess=
, and usual log sizes? other custom installs need tuning anyway

>I look forward to your thoughts and feedback on this proposal=2E
>
>Cheers,

indeed=2E we have large disks now=2E but we fill them all=2E i started wit=
h 1=2E2g one=2E was too small=2E needed to compress for space=2E now i have=
 12t=2E it's still too small=2E i compress for space=2E they make them up t=
o 22t nowadays=2E this is about 20000 times larger but still feels small=2E=
 how did this happen? we just did this to ourselves=2E data sizes have kept=
 up with storage and bandwidth=2E gamer might get 1gbit/s connection at hom=
e so (s)he only needs to wait for one hour to download new game=2E just as =
dialup user once did, wait for hours=2E or it could be photographer, graphi=
cs designer or architect who works=2E both cases still use compressed data =
as cpu and ram permits it and it saves a lot of time and space=2E now, it's=
 also related to servers as those things don't disappear to or appear from =
just thin air=2E they come from machines, some of them hopefully running fb=
sd, where admins wonder how to deal with large log sizes=2E they need them =
for audit purposes=2E or statistics=2E hardware allows, so they compress it=
=2E "write-only-read-never" data benefits from, eg, xz a lot=2E as others a=
lready have told

so yeah, from (only!) 25+ years of experience, i can confirm that humankin=
d has developed AND used everything at max=2E internet, first for military =
and educational uses, now for connecting washing machines=2E oh and, first =
hdd, state of art device then, can only store *part* of *compressed* photo =
now

now, this might not be related to default fbsd installs in common usage wh=
ere default base syslog creates tiny amount of data per week

but one of reasons was given how everything fits uncompressed nowdays=2E t=
o our disks and pipes=2E which it really doesn't



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E124EE62-15CA-48C7-BED1-BA21B5587EDF>