-
Notifications
You must be signed in to change notification settings - Fork 580
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
PID file removed after reload #6082
Comments
@Crunsher since you've change parts with the reload, any ideas? |
This definitely came with my change. Do you run SELinux? It could block the SIGUSR2 signal, possibly. |
@Crunsher SELinux is currently disabled. In the current version 2.8.1+414.gbb96b7742.2018.02.10+1.stretch-0 this issue is still active. |
I was unable to reproduce this on Centos 6 with (don't have a debian with custom sysvinit lying about)
Could you please run the reload with strace? |
|
We figured it out: You are using an old init script! |
This is not an old script, it's the init script for Debian based systems. We need to figure out the issue and fix it in Icinga/deb-icinga2#2 Edit: Please discuss there |
Sadly @lazyfrosch is right, this is not related to our changes. But since this is directly related to the debian init scripts, I'm closing this in favor of Icinga/deb-icinga2#4 |
Let's come back to the original problem, since I can't reproduce it with Debian stretch in sysV init mode. I've installed a fresh Debian stretch, rebooted it with sysV and configured Icinga 2 + IDO MySQL
Now let's see how restarting does:
@Crunsher Were you able to reproduce the error? @tmatthaeus what might be different in my setup compared to yours? |
@tmatthaeus have you had this problem on any other Stretch/sysV system? I can't reproduce on a fresh install. |
Okay so it doesn't look like OP is using sysV init, on closer inspection of his strace, it is systemd. So the initscript doesn't matter at all... To my surprise icinga2 is really dying during a reload with systemd:
This is currently only a problem with snapshots it seems. Yes the systemd file for Debian/Ubuntu differs, but I'm not sure why systemd is loosing the daemon here... |
I will switch the systemd script for Debian and Ubuntu to the one included with Icinga 2. Meanwhile we should make sure that the daemon updates the PID file correctly:
@Crunsher Could you have a look at this? |
Note: Affects current master starting with c418a96 and probably 2.8.2 then |
It doesn't happen in my custom debug build. So it is likely there might be some permission problem or similar. I'll see to this getting looked at before 2.8.2 hits.
|
Update: I thought this was a 2.8.2 issue but it's not 🤷♀️ |
I think the core problem with this issue is that we do not longer update the PID file on Can we discuss this tomorrow in person? I'd call |
I am not able to reproduce this:
We can easily replace our call to Exit with a call to Stop, though I'd like a way to test this first. |
The user is using systemd! systemd is expecting that the PID file is updated before the old daemon is exiting. We seem to no longer do that. So the "old" systemd unit fails. But we should make sure this still works, despite updating the unit file... |
With which, as I mentioned earlier, I'm unable to reproduce this either. |
After a reload of icinga2 there is no icinga2 process via /etc/init.d/icinga2 reload there is no icinga2 process anymore.
OS: Debian 9.3
Icinga Version:
Debug Logfile output:
After that no output anymore and no icinga2 process is running anymore.
The text was updated successfully, but these errors were encountered: