Hi All, after using this procedure some days I found some "bad" side effects. The most critical one for me was: FAI is killed and nobody knows that there is a device / server that has failed with installation. (This of course happens mainly if you are using FAI in an automated environment without somebody starting FAI manually)
After some thinking I came across a better (and more FAI fashion) solution which I of course want to share: if any of these invalid class combinations is defined my scripts checks for this and sets a new class: ERRORCLASS This class contains a single hook which skips all other tasks except the ones required for logging and correct FAI working. Based on this no bad action is performed during installation and I still have fully functional logging/reporting process. This is how the hook (action.ERRORCLASS.sh) looks like: *#! /bin/bash# version: 2019.11.01# author: Martin Kraemer, mk.mad...@gmail.com <mk.mad...@gmail.com># description: skip all tasks that change the system - keep only logging etc.##--write an error message for logging>&2 echo "$0: error: error within class configuration"#skiptask confdir#skiptask setup#skiptask defclass#skiptask defvar#skiptsk actionskiptask sysinfoskiptask inventory#skiptask install#skiptask dirinstall#skiptask softupdateskiptask partitionskiptask mountdisksskiptask extrbaseskiptask debconfskiptask repositoryskiptask updatebaseskiptask instsoftskiptask configureskiptask tests#skiptask finish#skiptask chboot#skiptask savelog#skiptask faiend* Am Sa., 26. Okt. 2019 um 22:02 Uhr schrieb Martin Krämer < mk.mad...@gmail.com>: > Hi Thomas, > > Thanks for your Investigation. > Yes - I was not able to find any related changes, too. > > Unfortunately as you already expected task_error does not end the full > process tree - only fai-class. > Never the less it helped me to find a good workaround which I use as > follows: > > >&2 echo "error: my critical message" > (kill -9 $PPID &) > task_error 123 > > Kind Regards > > Martin Krämer > > > On 26. Oct 2019, at 19:36, Thomas Lange <la...@informatik.uni-koeln.de> > wrote: > > > > Hi Martin, > > > > I can find no change that causes a change in the behaviour for that > > between 5.8.1 and the newest FAI version. As you pointed out an "exit X" > > in a .sh script causes fai-class to stop defining more classes. You > > can try to set an error code inside a .sh file by using a call to > > task_error xxx. I'm not sure if this will work, since fai-class runs in a > > subprocess. > > After looking into the code it seems that fai-class almost always ends > > with exit 0. > > > > -- > > regards Thomas >
