today i am starting holidays for two weeks, anyways i will try to explain you the real
world and why it doe not matter most of the time if a system boots 5 seconds faster if
it does this relieable
Am 20.06.2011 19:24, schrieb Lennart Poettering:
On Sun, 19.06.11 11:48, Reindl Harald (h.reindl(a)thelounge.net)
wrote:
> because there are a lot of services in LSB-Mode i filed for everyone a
> bugreport - sad to see that most packages are not F15 ready
systemd reads SysV/LSB init scripts as alternative configuration source if
no native unit file is available for a service.
it does not respect the start-order in tis case and fires services
up like a blind butcher - thi smay be enough for a simple desktop
but NOT if you have services depending on each other and started
means NOT ready for connections
means: it reads them and fires up braindead with only speed in mind
Due to that I'd very
much claim that those services in question are F15-ready. SysV/LSB
scripts are officially supported on F15 (and will continue to be
supported for a long time beyond that).
but this works only in your small world and on a desktop!
one example is a FULLY mysql-driven mail-server
* dbmail, postfix, dovecot-proxy, postfix-policyd depends on mysqld
* dovecot as imap proxy depends on dbmail-imapd
* postfix depends for sasl-auth on dovecot
this means explicitly that you have a defined start-order
started means in this case: ready for connections
processs running is not enough
below a well defined and since years working starting order
and now tell me how this will work with your blind "i fire up
every service as fast as i can to make boot 5 seconds faster"
when the result is nothing orks as expected?
additionally i have since years watchdog-scripts runnig whaich are restarting
services if something went wrong and detecting if "/sbin/service" for the
missing service is running or a reboot is requested to prevent blind fire up
them - this does not work any longer this way
so we come to the next major problem:
https://bugzilla.redhat.com/show_bug.cgi?id=714525
it does not really matter for the admin who and why fires up mysqld in
this case - if i say "systemctl stop mysqld" i mean this exactly so,
everytime in everycase and systemd has to respect this!
this start-order is hardly workd out, most services are compiled on an
own build-environment to prevent overwriting init-scripts and make
sure that exactly this start-order is active, everytime, 365 days
and 24 hours a day
[root@mail:~]$ ls /etc/rc3.d/
insgesamt 0
lrwxrwxrwx. 1 root root 19 2008-08-18 07:03 K10dc_server -> ../init.d/dc_server
lrwxrwxrwx 1 root root 19 2011-05-28 20:26 K10saslauthd -> ../init.d/saslauthd
lrwxrwxrwx. 1 root root 19 2008-08-18 07:03 K12dc_client -> ../init.d/dc_client
lrwxrwxrwx 1 root root 19 2011-05-28 20:27 K35denyhosts -> ../init.d/denyhosts
lrwxrwxrwx. 1 root root 20 2008-08-18 07:03 K50netconsole -> ../init.d/netconsole
lrwxrwxrwx 1 root root 16 2011-06-17 02:56 K50vnstat -> ../init.d/vnstat
lrwxrwxrwx 1 root root 15 2011-05-28 20:26 K75netfs -> ../init.d/netfs
lrwxrwxrwx 1 root root 17 2011-05-28 20:26 K75ntpdate -> ../init.d/ntpdate
lrwxrwxrwx 1 root root 20 2011-05-28 20:25 K85messagebus -> ../init.d/messagebus
lrwxrwxrwx 1 root root 15 2011-05-28 20:26 K89rdisc -> ../init.d/rdisc
lrwxrwxrwx 1 root root 18 2011-05-28 20:25 S08iptables -> ../init.d/iptables
lrwxrwxrwx 1 root root 17 2011-05-28 20:26 S10network -> ../init.d/network
lrwxrwxrwx 1 root root 16 2009-03-17 10:31 S11auditd -> ../init.d/auditd
lrwxrwxrwx 1 root root 17 2011-05-28 20:26 S12rsyslog -> ../init.d/rsyslog
lrwxrwxrwx 1 root root 17 2011-05-30 00:45 S14unbound -> ../init.d/unbound
lrwxrwxrwx 1 root root 19 2011-05-28 20:26 S26udev-post -> ../init.d/udev-post
lrwxrwxrwx 1 root root 14 2011-05-28 20:26 S55sshd -> ../init.d/sshd
lrwxrwxrwx 1 root root 14 2011-05-28 20:26 S58ntpd -> ../init.d/ntpd
lrwxrwxrwx 1 root root 16 2009-09-20 02:16 S64mysqld -> ../init.d/mysqld
lrwxrwxrwx 1 root root 32 2011-05-28 20:27 S79dbmail-postfix-policyd ->
../init.d/dbmail-postfix-policyd
lrwxrwxrwx 1 root root 18 2011-05-28 20:27 S79postgrey -> ../init.d/postgrey
lrwxrwxrwx 1 root root 18 2009-09-20 00:34 S80vmtoolsd -> ../init.d/vmtoolsd
lrwxrwxrwx 1 root root 22 2011-05-28 20:26 S81dbmail-imapd -> ../init.d/dbmail-imapd
lrwxrwxrwx 1 root root 22 2011-05-28 20:26 S81dbmail-lmtpd -> ../init.d/dbmail-lmtpd
lrwxrwxrwx 1 root root 22 2011-05-28 20:26 S81dbmail-pop3d -> ../init.d/dbmail-pop3d
lrwxrwxrwx 1 root root 26 2011-05-28 20:26 S81dbmail-timsieved ->
../init.d/dbmail-timsieved
lrwxrwxrwx 1 root root 17 2010-11-24 14:53 S82dovecot -> ../init.d/dovecot
lrwxrwxrwx 1 root root 17 2011-05-28 20:26 S83postfix -> ../init.d/postfix
lrwxrwxrwx 1 root root 15 2011-05-28 20:26 S85httpd -> ../init.d/httpd
lrwxrwxrwx 1 root root 15 2011-05-28 20:26 S90crond -> ../init.d/crond
lrwxrwxrwx 1 root root 11 2011-05-28 20:26 S99rc-local -> ../rc.local
Of course, we'd like to see more and more services moved over to
native
unit files and I greatly appreciate Johann's and Michał's work on this,
but that is work for F16, and should not be done on an already released
distribution.
then the distribution should not have been released with systemd