You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our build or server do not support creating core dumps when a process fails. The issues as I understand them are:
The OS defines the location and file name pattern for a core dump with /proc/sys/kernel/core_pattern. However, in the containuer it uses the definition from the host system and is marked as readonly. In (at least a ubuntu host) the definition for a core dump is normally:
|/usr/share/apport/apport %p %s %c %d %P %E
Which defines the apport app as the target with a bunch of other data for the app.
We do not install apport in the container so that all falls apart and core dumps are simply not recorded.
Since we do not set ulimit. starting the container with the option --ulimit core=0 does officially disable core dumps. An alternative is apparently to set kernel.core_patther=/dev/null
Not sure if we need core dumps. I need the capability right now but that is temporary trying to figure out a specific problem and I might be able to do the same thing with a vull VM
I am guessing that apport is probably not the way we want to generate them if we do enable core dumps.
The text was updated successfully, but these errors were encountered:
Our build or server do not support creating core dumps when a process fails. The issues as I understand them are:
|/usr/share/apport/apport %p %s %c %d %P %E
Which defines the apport app as the target with a bunch of other data for the app.
We do not install apport in the container so that all falls apart and core dumps are simply not recorded.
Since we do not set ulimit. starting the container with the option --ulimit core=0 does officially disable core dumps. An alternative is apparently to set kernel.core_patther=/dev/null
The text was updated successfully, but these errors were encountered: