9:52 a.m.
so a couple of weeks ago,
I did a dnf update, and got ~ 19% reduction due to DRPM
that was by far the highest return on the effort Id seen.
previous experience for me was 1-4%.
I should have grabbed a "screenshot" and sent it here.
Whats the "average" experience here ?
are there statistics ?
do *noarch 's do better ?
how often do you see some version of "find/download failed" ?
(it sometimes appears to be a cached-version problem,
particularly if the update is older than a few days)
10:03 a.m.
On Wed, 2 Feb 2022 08:52:17 -0700
jim.cromie(a)gmail.com wrote:
Whats the "average" experience here ?
My average for years was about 1%, and half of those would say "checksums
don't match" and download the full rpm anyway. I turned off drpm
a while back because it was so useless.
3:16 a.m.
Tom Horsley:
My average for years was about 1%, and half of those would say
"checksums don't match" and download the full rpm anyway. I turned
off drpm a while back because it was so useless.
I suspect it's only going to really advantage people that do daily
updates, or update almost every time there's a new version released.
If you only do updates one a week, or less often, there may be several
updates in the meantime, and the delta mayn't be able to span that many
releases.
I used to see some surprisingly small DRPM downloads. It's no real
surprise when you consider that a bug fix to a package may probably
only be a small tweak. I used to razz a Windows friend about very
quickly downloading lots of updates of only a few kilobytes, and doing
it in the background, without stuffing up what I was doing at the time,
and rarely needing to reboot. He was re-installing his entire OS
almost once a week.
--
uname -rsvp
Linux 3.10.0-1160.53.1.el7.x86_64 #1 SMP Fri Jan 14 13:59:45 UTC 2022 x86_64
Boilerplate: All unexpected mail to my mailbox is automatically deleted.
I will only get to see the messages that are posted to the mailing list.
8:23 a.m.
On Wed Feb02'22 11:03:46AM, Tom Horsley wrote:
From: Tom Horsley <horsley1953(a)gmail.com>
Date: Wed, 2 Feb 2022 11:03:46 -0500
To: users(a)lists.fedoraproject.org
Reply-To: Community support for Fedora users <users(a)lists.fedoraproject.org>
Subject: Re: DRPM performance ?
On Wed, 2 Feb 2022 08:52:17 -0700
jim.cromie(a)gmail.com wrote:
> Whats the "average" experience here ?
My average for years was about 1%, and half of those would say "checksums
don't match" and download the full rpm anyway. I turned off drpm
a while back because it was so useless.
I am curious, how does one turn off drpm?
Best wishes,
Ranjan
8:31 a.m.
On Sat, 19 Feb 2022 08:23:02 -0600
Ranjan Maitra wrote:
I am curious, how does one turn off drpm?
In /etc/dnf/dnf.conf add this line in [main]
deltarpm=False
It is in the dnf docs (surprisingly :-).
Another good one is
color=never
Which keeps my terminal from looking like an explosion in a paint factory :-).
11:24 p.m.
On Sat Feb19'22 09:31:03AM, Tom Horsley wrote:
From: Tom Horsley <horsley1953(a)gmail.com>
Date: Sat, 19 Feb 2022 09:31:03 -0500
To: users(a)lists.fedoraproject.org
Reply-To: Community support for Fedora users <users(a)lists.fedoraproject.org>
Subject: Re: DRPM performance ?
On Sat, 19 Feb 2022 08:23:02 -0600
Ranjan Maitra wrote:
> I am curious, how does one turn off drpm?
In /etc/dnf/dnf.conf add this line in [main]
deltarpm=False
It is in the dnf docs (surprisingly :-).
Another good one is
color=never
Which keeps my terminal from looking like an explosion in a paint factory :-).
Thanks for this, really handy to know.
Ranjan
7:45 a.m.
On Mon, 21 Feb 2022 10:24:29 +0000
Patrick O'Callaghan wrote:
I didn't even know dnf had a colour (or color) option. It seems
to be
off by default.
Certainly wasn't off by default when strange colorful output
started appearing in my terminal when running dnf. That's why
I looked up how to make it stop.
1:09 p.m.
On Wed, Feb 2, 2022 at 8:52 AM <jim.cromie(a)gmail.com> wrote:
so a couple of weeks ago,
I did a dnf update, and got ~ 19% reduction due to DRPM
just the other day, I saved 18%, so it does happen semi-regularly
FWIW, I wouldnt mind seeing %saved over the DPRM subset,
not just %saved vs the whole download.
9:55 a.m.
On Wed, Feb 2, 2022 at 8:52 AM <jim.cromie(a)gmail.com> wrote:
so a couple of weeks ago,
I did a dnf update, and got ~ 19% reduction due to DRPM
that was by far the highest return on the effort Id seen.
previous experience for me was 1-4%.
I should have grabbed a "screenshot" and sent it here.
Whats the "average" experience here ?
are there statistics ?
do *noarch 's do better ?
how often do you see some version of "find/download failed" ?
(it sometimes appears to be a cached-version problem,
particularly if the update is older than a few days)
One more data-point:
I got 36% saved this morning.
7/10 packages were drpm'd
that seems worth having, esp for the servers.
Last metadata expiration check: 0:09:04 ago on Mon 11 Apr 2022 08:27:41 AM MDT.
Dependencies resolved.
================================================================================
Package Arch Version Repository Size
================================================================================
Upgrading:
selinux-policy noarch 35.16-1.fc35 updates 68 k
selinux-policy-targeted noarch 35.16-1.fc35 updates 6.3 M
systemd x86_64 249.11-1.fc35 updates 4.1 M
systemd-container x86_64 249.11-1.fc35 updates 523 k
systemd-libs x86_64 249.11-1.fc35 updates 603 k
systemd-networkd x86_64 249.11-1.fc35 updates 523 k
systemd-oomd-defaults noarch 249.11-1.fc35 updates 24 k
systemd-pam x86_64 249.11-1.fc35 updates 323 k
systemd-resolved x86_64 249.11-1.fc35 updates 266 k
systemd-udev x86_64 249.11-1.fc35 updates 1.8 M
Transaction Summary
================================================================================
Upgrade 10 Packages
Total download size: 14 M
Is this ok [y/N]: y
Downloading Packages:
(1/10): systemd-libs-249.9-1.fc35_249.11-1.fc35 93 kB/s | 104 kB 00:01
(2/10): systemd-container-249.9-1.fc35_249.11-1 194 kB/s | 262 kB 00:01
(3/10): systemd-pam-249.9-1.fc35_249.11-1.fc35. 272 kB/s | 83 kB 00:00
[DRPM 1/7] systemd-libs-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
[DRPM 2/7] systemd-container-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
(4/10): systemd-networkd-249.9-1.fc35_249.11-1. 246 kB/s | 151 kB 00:00
(5/10): systemd-resolved-249.9-1.fc35_249.11-1. 351 kB/s | 87 kB 00:00
[DRPM 3/7] systemd-pam-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
(6/10): selinux-policy-35.16-1.fc35.noarch.rpm 357 kB/s | 68 kB 00:00
(7/10): systemd-udev-249.9-1.fc35_249.11-1.fc35 375 kB/s | 477 kB 00:01
[DRPM 4/7] systemd-networkd-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
[DRPM 5/7] systemd-resolved-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
(8/10): systemd-oomd-defaults-249.11-1.fc35.noa 351 kB/s | 24 kB 00:00
(9/10): systemd-249.9-1.fc35_249.11-1.fc35.x86_ 379 kB/s | 1.7 MB 00:04
(10/10): selinux-policy-targeted-35.16-1.fc35.n 996 kB/s | 6.3 MB 00:06
[DRPM 6/7] systemd-udev-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
[DRPM 7/7] systemd-249.9-1.fc35_249.11-1.fc35.x86_64.drpm: done
--------------------------------------------------------------------------------
Total 1.0 MB/s | 9.2 MB 00:09
Delta RPMs reduced 14.4 MB of updates to 9.2 MB (36.0% saved)
12:24 p.m.
On 4/11/22 07:55, jim.cromie(a)gmail.com wrote:
> One more data-point:
I got 36% saved this morning.
7/10 packages were drpm'd
that seems worth having, esp for the servers.
Total 1.0 MB/s | 9.2 MB 00:09
Delta RPMs reduced 14.4 MB of updates to 9.2 MB (36.0% saved)
That sounds great until you see that it was 5MB that you saved. The
deltarpm metadata is bigger than that.
743
days inactive
811
days old
10 comments
6 participants
participants (6)
-
jim.cromie＠gmail.com -
Patrick O'Callaghan -
Ranjan Maitra -
Samuel Sieb -
Tim -
Tom Horsley