I re-installed the former package content, rebooted the system and gave it some more memory to get rid of any concerns in that regard (from the perf data).
The diff of the actual content before/after was all reasonable (new times, different package counts, but otherwise the same)
Consumption wise we have:
# Before fixes
real 1m11.731s
us sy id wa st
71 20 9 0 0
At a very high summary:
- delay cut down by 2/3 of the initial duration
- consumption reduced by 71% (duration) and by 26% (consumption while running).
This is multiplicative so for the example we have an saving of ~81.27% of cpu cycles
I re-installed the former package content, rebooted the system and gave it some more memory to get rid of any concerns in that regard (from the perf data).
The diff of the actual content before/after was all reasonable (new times, different package counts, but otherwise the same)
Consumption wise we have:
# Before fixes
real 1m11.731s
us sy id wa st
71 20 9 0 0
59.16% landscape-sysin
14.23% swapper
6.16% sshd
5.78% lsb_release
3.35% apt-config
0.86% dbus-daemon
0.80% systemd
0.59% gdbus
0.56% dpkg
0.48% systemd-logind
# After fixes
real 0m21.257s
us sy id wa st
32 34 34 0 0
42.00% swapper
22.45% sshd
2.94% dbus-daemon
2.79% systemd
2.08% gdbus
2.07% grep
1.88% run-parts
1.58% systemd-logind
1.46% find
1.23% bash
1.22% update-motd-fsc
1.18% systemd-journal
1.13% cat
That is good.
Ready to open MRs for this.
At a very high summary:
- delay cut down by 2/3 of the initial duration
- consumption reduced by 71% (duration) and by 26% (consumption while running).
This is multiplicative so for the example we have an saving of ~81.27% of cpu cycles