apache2 starts properly (/etc/init.d/apache2 start) but will NOT start when booting, even though it is properly configured in yast System->System Services (Runlevel)
yast says everything is ok, but the service won’t start unless I manually start it.
I have another application that does the same thing, but I think the issue preventing apache2 from starting will help me with the other issue.
I’ve worked with a number of versions on SUSE and I’ve never seen this happen before. It’s driving me crazy (a short trip) so if someone could point me in the right direction I’d really appreciate it.
Obviously others are not seeing this issue, but I can’t see any reasonable explanation!
Any debug help would be good, but I’ve worked on this quite a while… and I’ve done a lot of research, so I think I’m beyond the obvious but… who knows…??
I can start apache2 in yast, but apache2 fails to start when the server is rebooted.
On 2012-10-07 02:06, lashworth wrote:
>
> version 12.2 (latest) 64 bit
Oops, did not see the title. So, systemd. You might try booting with systemv instead.
I think I recall reading something similar, but I’m not sure where.
–
Cheers / Saludos,
Carlos E. R.
(from 12.1 x86_64 “Asparagus” at Telcontar)
Does “systemctl --failed” show the service as failed? What “systemctl status apache2.service” show? Finally there could be some hints in “systemd-journalctl”.
I removed NetworkManager and the apache2 socket bindings all work now… Why it made a difference I have no clue. Love to learn if anyone can explain why…
On 2012-10-08 20:36, lashworth wrote:
>
> I removed NetworkManager and the apache2 socket bindings all work now…
> Why it made a difference I have no clue. Love to learn if anyone can
> explain why…
Of course it does! The network has to be running completely before apache starts.
–
Cheers / Saludos,
Carlos E. R.
(from 12.1 x86_64 “Asparagus” at Telcontar)
It retries. On my laptop I run Networkmanager and a LAMP server. Works fine, never had any issues. Yes, Yast http-server warns I’m running networkmanager, but it all works. The thing you have to do, is install the LAMP pattern, then enter Yast - Networkservices - HTTP Server, have it starting at boot, open firewall port, done. If you want mysql to start as well, you’ll have to activate it through the runlevel editor. If you start apache2 this way, it works. Not just on my own laptop.
I was still having port binding issues with other java based applications that I use on our server, so out of desperation I put NetworkManager back in the lineup, and now everything works. The problem is that I don’t think I learned anything from this exercise. It works which is great, but I’d feel better if I understood what was going on.
I’m going to be upgrading a couple more servers shortly, and I don’t see what I did that caused the issue. I hope I don’t go through that again.
I appreciate your help. Sometimes a reality check, and a reminder of the tools available to locate issues is really a good assist.
Helps you break the loop.
Thanks.
btw: Is there an online set of suggestions on how to debug openSUSE Linux installs that you really like? Most of the hints are excessively focused on specific issues.
A general set of helps applicable to openSUSE would be a great help.
Especially since there seems to be significant differences between Linux builds.
Traditional network startup script configures interfaces synchronously. So one can be more or less sure that when script has finished, interfaces are indeed up. NM startup script simply starts NetworkManager which configures interfaces asynchronously. So after NM startup script has finished, nothing can be said about network state yet. If you depend on interfaces being available you can set NM_ONLINE_TIMEOUT in /etc/sysconfig/network/config. This defines upper limit for how long system will wait after starting NetworkManager for network being up. Default is 0 as far as I can tell, so no waiting.
lashworth wrote:
> btw: Is there an online set of suggestions on how to debug openSUSE
> Linux installs that you really like?
The single thing that most surprises me on these forums, and which
occurred in this case, is that somebody posts a question without posting
evidence from the relevant logs. And then nobody even suggests that they
should do so!
Almost the first thing to do with almost all problems, IMHO, is to find
out what the relevant logs might be and go look in them.
This would explain why your suggestion to use sysvinit rather than systemd would work. I feel a lot better now. It looks like a bug that will get fixed right away, and I know what to do until it is addressed. It does seem logical that both start up systems should wait until interfaces are completely configured before continuing, and that the logic of the sysvinit and systemd start up processes should be the same on this issue.