On 2026-09-24 10:31, Thorsten Kukuk via Postfix-users wrote:
You are doing something wrong here. It doesn't matter if your
container runs on Alpine or a containerized systemdD distro, why
should that matter?
And which configuration issues? Neither docker nor podman nor k8s
depends on systemD. Except you have to create a systemd service for
them, but at least podman does not for you automatically.
Or do you mean configuring postfix itself? Yes, this depends on the
author of the postfix container.
My postfix container needs 164MB additional disk space (though the
base image is shared with many other containers, so in reality it's
much less) and at runtime it requires the same amount of memory as if
I would run it directly on the system, at least if the libraries are
identical to the host. If they differ, you need of course more memory
because the kernel cannot share the libraries with other processes.
Thorsten
careful: before telling someone that they are doing something wrong,
you may want to check your own assumption. specifically on what
containerization tech is being used, and possibly confusing RAM vs disk
space.
$ incus image list images: | grep amd64 | grep -i suse | grep cloud |
grep -i container
| opensuse/16.0/cloud (1 more) | 1a19f6e590d0 | yes |
Opensuse 16.0 amd64 (20260924_04:20) | x86_64 | CONTAINER
| 77.11MiB | 2026/09/23 20:00 EDT |
| opensuse/tumbleweed/cloud (1 more) | 3cf5befc77fa | yes |
Opensuse tumbleweed amd64 (20260924_04:20) | x86_64 | CONTAINER
| 79.04MiB | 2026/09/23 20:00 EDT |
$ incus image list images: | grep amd64 | grep -i ubuntu | grep cloud |
grep -i container
| ubuntu/26.04/cloud (3 more) | 4b24b84e202d | yes |
Ubuntu resolute amd64 (20260924_07:42) | x86_64 | CONTAINER
| 165.43MiB | 2026/09/23 20:00 EDT |
| ubuntu/jammy/cloud (3 more) | fe0c2856d136 | yes |
Ubuntu jammy amd64 (20260924_07:42) | x86_64 | CONTAINER
| 156.10MiB | 2026/09/23 20:00 EDT |
| ubuntu/noble/cloud (3 more) | b39d3d56c307 | yes |
Ubuntu noble amd64 (20260924_07:42) | x86_64 | CONTAINER
| 161.52MiB | 2026/09/23 20:00 EDT |
$ incus image list images: | grep amd64 | grep -i alpine | grep cloud |
grep -i container
| alpine/3.21/cloud (1 more) | 850c68ae2403 | yes |
Alpine 3.21 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 19.25MiB | 2026/09/16 20:00 EDT |
| alpine/3.21/tinycloud (1 more) | 5092ecae3791 | yes |
Alpine 3.21 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 4.78MiB | 2026/09/16 20:00 EDT |
| alpine/3.22/cloud (1 more) | 22902182a7ac | yes |
Alpine 3.22 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 19.32MiB | 2026/09/16 20:00 EDT |
| alpine/3.22/tinycloud (1 more) | 74370825e9f6 | yes |
Alpine 3.22 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 5.20MiB | 2026/09/16 20:00 EDT |
| alpine/3.23/cloud (1 more) | 851b8adb2b3d | yes |
Alpine 3.23 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 19.44MiB | 2026/09/16 20:00 EDT |
| alpine/3.23/tinycloud (1 more) | 8997201b86d6 | yes |
Alpine 3.23 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 5.27MiB | 2026/09/16 20:00 EDT |
| alpine/3.24/cloud (1 more) | c32400642903 | yes |
Alpine 3.24 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 20.22MiB | 2026/09/16 20:00 EDT |
| alpine/3.24/tinycloud (1 more) | 56cdd2b12244 | yes |
Alpine 3.24 amd64 (20260917_13:00) | x86_64 | CONTAINER
| 5.21MiB | 2026/09/16 20:00 EDT |
| alpine/edge/cloud (1 more) | bc9385a07c49 | yes |
Alpine edge amd64 (20260917_13:00) | x86_64 | CONTAINER
| 20.76MiB | 2026/09/16 20:00 EDT |
| alpine/edge/tinycloud (1 more) | 25bab84574bc | yes |
Alpine edge amd64 (20260917_13:00) | x86_64 | CONTAINER
| 5.33MiB | 2026/09/16 20:00 EDT |
$ incus launch images:1a19f6e590d0 susetest
Launching susetext
$ incus launch images:4b24b84e202d ubutest
Launching ubutest
$ incus launch images:c32400642903 alptest
Launching alptest
$ for VM in `incus list --columns n --format csv | grep test` ; do echo
${VM} ; incus info ${VM} | grep "Memory (" ; done
alptest
Memory (current): 12.39MiB
susetest
Memory (current): 47.80MiB
ubutest
Memory (current): 86.25MiB
these are just immediately after launching, and only base O/S. my
practical experience of approximately 300MB difference is on live
production containers, apples to apples and oranges to oranges (e.g. a
MariaDB database on Ubuntu vs a MariaDB database on Alpine). I included
Suse above because I see you are a SuSE engineer. I have not had any
practical experience with SuSE in recent times, so maybe you guys have
some extra magic sauce that Ubuntu does not.
and a few hours of *in*activity later:
$ for VM in `incus list --columns n --format csv | grep test` ; do echo
${VM} ; incus info ${VM} | grep "Memory (" ; done
alptest
Memory (current): 12.65MiB
susetest
Memory (current): 47.85MiB
ubutest
Memory (current): 205.89MiB
so maybe the culprit is something that the wannabe Microsoft of the
Linux distros does. Nevertheless: o/s diversity is as important as
biodiversity; and if all o/s become a copycat of the 800lbs gorilla, all
with the same init package, we are one AIbola away from disaster.
while once RAM-AI-geddon is over most use cases will be able to afford
this overhead waste, it is not a good reason to tolerate it.
Yuv
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
