Several 12.2 programs startup problems.

I’m very disappointed about the low quality of systemd startup configuration.
I moved an 11.4 web/mail/dns server into 3 different machines, all with new installed 12.2.
In all 3 server I found problems with packages not starting or stopping or reloading correctly.

  1. Apache2 starts but goes on signaling errors:
    systemd[1]: PID file /var/run/httpd2.pid not readable (yet?) after start.

  2. Fail2ban says it does not start but it starts and does not stop until I kill it:
    Oct 29 21:42:57 rohan fail2ban[14866]: Starting fail2ban …failedOct 29 21:42:57 rohan systemd[1]: fail2ban.service: control process exited, code=exited status=7
    Oct 29 21:42:57 rohan systemd[1]: Unit fail2ban.service entered failed state.

  3. Postgrey does not start at all:
    Oct 27 19:01:08 rohan postgrey[17022]: Starting postgrey daemondefined(@array) is deprecated at /usr/lib/perl5/vendor_perl/5.16.0/Net/Server.pm line 211.
    Oct 27 19:01:08 rohan postgrey[17022]: (Maybe you should just omit the defined()?)
    Oct 27 19:01:08 rohan postgrey[17022]: …failed
    Oct 27 19:01:08 rohan systemd[1]: postgrey.service: control process exited, code=exited status=7
    Oct 27 19:01:08 rohan systemd[1]: Unit postgrey.service entered failed state.

These are only the “important” packages I noticed to have problems, I did not look at minor ones.
All bugs are already signaled, but no solution are available.
It’s the first time I see a OpenSuse release to have generalized problems making some packages unusable.
At least the startup of packages should be tested well in a distro.

Hi welcome here,

I moved an 11.4 web/mail/dns server into 3 different machines

How? What do you mean by this?

Ad. 1: I have no issues with apache2 at all. This looks like a permission problem, which brings me back to the first questions

Somehow I get the idea that you simply copied over stuff from 11.4 into a 12.2 install. That would explain a lot.

12.2 is described as being rockstable, personal experiences are no different. My gut feeling says there’s something wrong in the methods used, but I hope I’m wrong, also hope this is not on production machines

I did a fresh installation under VmWare of 3 new virtual machines. I did not copy anything from old system. Also the config files where rewritten completely.
After 10 years of using OpenSuse, i did the error to start in production the mail server, it works, but some of the packages must be started stopped manually. Only postgrey does not work at all.

I found reports about this problem already reported, but no solutions:
[Bug 782364] New: openSUSE:12.2/postgrey: Bug](http://lists.opensuse.org/opensuse-bugs/2012-09/msg04584.html)

I cannot believe that 12.2 could go out without checking if a package is able to start or not. In this case it is a perl compatibility error, no configuration issues.

Did you test it before release?

I admit I did not check all packages, only major ones. This was indeed my error. I was too confident … now it’s too late to go back to 11.4.
Systems are up and running, but I suspect a generalized problem in systemd, not beeing able to check the status of running process and to manage them correctly.
I have to check them manually whit ps -ef, finding the pid and eventually killing them if systemctl does not do it correctly.
Yes, fault is mostly mine. I hope these problems would be addressed soon …

On 2012-10-31 06:36, fmalfatto wrote:
> I cannot believe that 12.2 could go out without checking if a package
> is able to start or not. In this case it is a perl compatibility error,
> no configuration issues.

As openSUSE is a community effort, it is your responsibility to test those things before release and
report.

Yes, I’m sure that systemd has problems, many. I’m no fan of it. If you want that the people that
put systemd in openSUSE read of your problem, report in the factory mail list >:-)


Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

The only “problem” that can be remotely attributed to systemd is the warning about apache and it is what it says - warning. And the right one. How can systemd be responsible for failures in programs in two other cases?

On 2012-10-31 17:36, arvidjaar wrote:
>
> robin_listas;2500139 Wrote:
>> I’m sure that systemd has problems
>
> The only “problem” that can be remotely attributed to systemd is the
> warning about apache and it is what it says - warning. And the right
> one. How can systemd be responsible for failures in programs in two
> other cases?

Well, try booting in systemv and find out.


Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

Actually it is running, even though it says “failed”

opensuse:~ # /etc/init.d/fail2ban start
redirecting to systemctl
Job failed. See system journal and 'systemctl status' for details.
opensuse:~ # /etc/init.d/fail2ban status
redirecting to systemctl
fail2ban.service - LSB: Bans IPs with too many authentication failures
      Loaded: loaded (/etc/init.d/fail2ban)
      Active: failed (Result: exit-code) since Thu, 01 Nov 2012 07:09:15 +0400; 2s ago
     Process: 11456 ExecStart=/etc/init.d/fail2ban start (code=exited, status=7/NOTRUNNING)
      CGroup: name=systemd:/system/fail2ban.service
          └ 11466 /usr/bin/python /usr/bin/fail2ban-server -b -s /va...

Nov 01 07:09:15 opensuse.site fail2ban[11456]: Starting fail2ban ..failed
opensuse:~ # export SYSTEMD_NO_WRAP=1
opensuse:~ # /etc/init.d/fail2ban status
Checking for service fail2ban                                        running
opensuse:~ # 

On 2012-11-01 04:26, arvidjaar wrote:
>
> robin_listas;2500268 Wrote:
>>
>> Well, try booting in systemv and find out.
>>
>
> Actually it is running, even though it says “failed”

Ok, but do try to find out the difference.


Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

opensuse:~ # /etc/init.d/fail2ban start
Starting fail2ban                                                    failed
opensuse:~ # ps -efwww | grep fail2ban
root      2653     1  0 20:18 ?        00:00:00 /usr/bin/python /usr/bin/fail2ban-server -b -s /var/run/fail2ban/fail2ban.sock

And where exactly is the difference?

On 2012-11-01 17:26, arvidjaar wrote:
>
> robin_listas;2500439 Wrote:
>> Ok, but do try to find out the difference.
>
>
> Code:
> --------------------
> opensuse:~ # /etc/init.d/fail2ban start
> Starting fail2ban failed
> opensuse:~ # ps -efwww | grep fail2ban
> root 2653 1 0 20:18 ? 00:00:00 /usr/bin/python /usr/bin/fail2ban-server -b -s /var/run/fail2ban/fail2ban.sock
> --------------------
>
> And where exactly is the difference?

You have to reboot in systemv mode. Did you?


Cheers / Saludos,

Carlos E. R.
(from 11.4 x86_64 “Celadon” (Minas Tirith))

init.d script for fail2ban is broken and can return success only by accident due to race condition. I do not need to reboot into systemv mode to see this.

I was able to fix the posgrey-perl problem by changing /usr/lib/perl5/vendor_perl/5.16.0/Net/Server.pm at line 211
from
$self->process_args( @ARGV, $template ) if defined @ARGV;to
$self->process_args( @ARGV, $template ) ;

Now also postgrey is afllen into systemd problems: starting but not recognized, not stopping, not refreshing):

root@rohan:/root # ps -ef | grep postgrey
root 4944 3821 0 11:15 pts/0 00:00:00 grep --color=auto postgrey

root@rohan:/root # systemctl start postgrey.service
Job failed. See system journal and ‘systemctl status’ for details.

root@rohan:/root # ps -ef | grep postgrey
postgrey 4955 1 0 11:16 ? 00:00:00 /usr/sbin/postgrey -d --unix=/var/spool/postfix/postgrey/socket --auto-whitelist-clients
root 4957 3821 0 11:16 pts/0 00:00:00 grep --color=auto postgrey

root@rohan:/root # systemctl stop postgrey.service

root@rohan:/root # ps -ef | grep postgrey
postgrey 4955 1 0 11:16 ? 00:00:00 /usr/sbin/postgrey -d --unix=/var/spool/postfix/postgrey/socket --auto-whitelist-clients
root 4979 3821 0 11:16 pts/0 00:00:00 grep --color=auto postgrey

I don’t want uninstall systemd (there are many warnings about doing that).
I simply think that such an important daemon should be at least doing the most part of its duty, before to go into a delivered system!

Thread resurrection :
i met the 12.2 fail2ban problem also, because i didn’t receive the start and stop mails from fail2ban. And i realized that some init infos are missing in the beginning of the /etc/init.d/fail2ban script : $postfix should be added in the Required-Start and in the Required-Stop sections . That’s a bug, i think. Was it already reported ? If not, where shall i report it ?
Nevertheless, using the command

systemctl status fail2ban.service

still reports failed whereas it is running and properly baning abusive ssh logging attempts. That is also a bug. I’m sure suse people are aware of this bug. Is there a bug report somewhere i could vote for ?

openSUSE:Submitting bug reports - openSUSE

possibly this?

https://bugzilla.novell.com/show_bug.cgi?id=790557

OK. Applied the change suggested by Fred. It works now.
Couldn’t vote for the bug.
I don’t why the change still hasn’t been made in the official updates.

Thank you anyway