On 5/10/2011 6:46 AM, Steffen Barszus wrote:
So the discussion should be on how to evaluate systemd , and set a
number of criterias to benchmark both. Then the better one should be
planned for slow migration.
"Look its new and it has bells and whistles lets move to that" is not
a valid argument for moving to a new init.
I agree, so let's see if we can get the ball rolling in that direction.
Some of the shortcomings I see with upstart that systemd sounds like it
addresses are:
1) The ability to drop back to single user mode for system
administration. Upstart can not make sure that all processes started
during multi user mode that are not supposed to be running in single
user mode are actually killed.
2) The ability to serialize state and re-execute itself so it can run
in the initramfs and then hand off to the real system.
3) The ability to figure out why a given service runs or does not run
when it does or should. Right now Upstart relies on comments being
added to the config files to give the admin clues about this, but that
is inherently unreliable and still difficult to figure out.
4) The increased parallelism systemd offers by running services at the
same time as those they "depend" on by having systemd create the socket
is not something upstart currently can do. I am not sure if it can be
extended to do this or not.
--
ubuntu-devel mailing list
[email protected]
Modify settings or unsubscribe at:
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
