Hi fellows,

I'm currently designing the logging system for our multi CPU system.
Up to now, I planned the following (original idea):
- start modified slogger on Node A (logs also the sending nodename, so we know which node wrote to the slogger's /dev/
slog)
- set symlink on Node B from /dev/slog to /net/A/dev/slog
- use slog() normally on both nodes (A and B)
This worked so far.

Now the problem/the question is the following:
when the "log collecting" node A is not availble, I want node B to log locally. So I've modified the original design 
described above. I'd like to use the QNX FS feature that files can be "stacked" upon each other, i.e. if one resource 
manager dies, the dev-file below that has been mounted below it appears and will be switched to seamlessly. So for node 
B my idea was to:
* start a local, modified slogger that creates /net/A/dev/slog with _RESMGR_FLAG_AFTER 
* start IP/QNet
What I expect this to do:
- as long as QNet hasnt come up, node B logs locally to its slogger (to /dev/slog, which points to /net/A/dev/slog, 
which is actually the local slogger)
- once QNet comes up, the local slogger is hidden by the QNet directory /net/A, and the remote slogger is used
- in case QNet dies down, logs get collected locally

What do you think, will this work? Of course I could do this by modifying the local slogger as well, but then I'd have 
_yet another_ special slogger version... so I'd rather try to do it with the onboard tools.

Greetings & Thanks ("Cheers"),
 Marc Roessler