Dave Brondsema wrote: > David Crossley wrote: [snip] > >Yes, i had already thought of that too. I was wondering if > >when forrestbot starts we can create a flag file in the > >forrestbot-logs directory, which gets removed at the end > >of the process. I have started a coding experiment. > >We would probably also need a way to check if there was > >a stale flag file lying around. > > Hrm, that is a tricky issue. The "right way" would be to see if the > forrest process is still running in the system (But what if you had > multiple users with write-access to a remotely mounted directory. > That'd mean different systems with only the log dir in common). But > doing system-level work would be non-portable anyway. So how to know > the difference between a long-running forrestbot and one that died > unexpectedly?
I wondered if adding a parameter to the project configuration file which provided an indication of how long it should take. Mmm, that sounds clunky and non-portable too. > In the forrestbot webapp, it says "running" if the log file is > incomplete (no BUILD SUCCESFUL). But if the log file's date (last > updated date, IIRC) is more than a couple minutes old it says "failed" > instead. Perhaps something like that would work. In the current code, it is "less than 60 seconds" which is way too short for something like cocoon-trunk. A releated issue is that the logfile only gets created when forrest actually starts. With cocoon-trunk the 'svn update' takes a long time so the logfile isn't yet updated. That should be easy to fix. --David
