1. Don't set a custom handler for SIGCHLD (in the umbrella process)
as that handler doesn't actually handle SIGCHLD anymore
2. Don't reset the SIGCHLD handler (in the worker process)
as there's nothing to reset anymore due to the above change
3. Don't block SIGCHLD across fork(2) as its handler doesn't change anymore
due to the above changes
Before:
On SIGCHLD from the forked worker the umbrella process sets a failure flag.
StartUnixWorker() recognises that and does waitpid(), failure message, etc..
On OpenBSD we can't tell the signal source, so we always set the failure flag.
That's not how our IPC shall work, that breaks the IPC sooner or later.
After:
No SIGCHLD handling and no failure flag setting.
Instead StartUnixWorker()'s wait loop uses waitpid(x,y,WNOHANG)
to avoid false positives while watching the forked worker.
As silent now no longer only controls the generation of log messages, a better
name is required. This changes its name, inverts its value to reflect the new
name and adds a documentation comment.
Since commit d9010c7b9faaec137f3e195b370edbb406c37d76, ActivateItems no
longer uses the WorkQueue upq to perform tasks but instead performs
these locally. One instance of `upq.Join()`/`upq.HasExceptions()`
remained in the function, but I believe this was just missed when
removing the `upq.Enqueue()` call just before.
This commit removes the corresponding parameter and updates all call
sites accordingly.
Introduce `-DInternal.DebugWorkerDelay=120` and sleep inside `RunWorker()`.
Rationale: With 2.11 we've introduced a real umbrella process where the
main process is spawned as child fork. Running icinga2 in foreground on
a macOS shell with LLDB will now exit, and not follow the child process.
LLDB doesn't support `follow-fork-mode child` like GDB and therefore we
need to:
- Print the child process PID
- Sleep for X seconds to allow the developer to run `lldb -p <PID>`, set breakpoints, etc.
This commit also documents all available debug build enabled internal constants.
Icinga processes check si_pid on IPC signals to ensure that the signal
comes from an expected process. OpenBSD doesn't currently fill in si_pid
in siginfo (leaving it 0) so signals never pass this check, resulting
in startup failing to complete (the worker doesn't see the umbrella's
signal and vice-versa) - issue #7505.
Relax the requirement so that either 0 or the wanted pid is accepted.
This partially reverts #7150 and avoids exceptions
inside the flow. Each time an empty active stage
is detected, Icinga tries to repair it from the
the given directory tree.
Also, the code now takes into account that it should
create the package storage on startup, whether within
the API object, or if disabled, inside the application.
Caching the active stages for packages in memory
only is in effect with the API feature being enabled.
This is useful for other deployed config packages,
not only the internal one.
fixes#7173
refs #7150
refs #7119fixes#6959
