wiki:WatchDog

Version 5 (modified by davea, 16 years ago) (diff)

--

Watchdogs

T(DesignDocument)?

A watchdog is a mechanism for detecting system states (e.g. full filesystems, database failures, etc.) that require immediate attention by project staff. Typically the desired response to such a condition is to notify a pager, sending a short text description.

BOINC provides a framework for defining watchdogs:

  • A set of watchdog scripts are run as periodic tasks. Each script checks for an error condition, and present, it appends a descriptive line to an error log file. An example is wd_nresults_changing.php, which makes sure that the number of results changes.
  • The script watchdogs.php, also run as a periodic task, scans the error log file. If it has been updated since the last run, it sends email to a set of recipients, containing the last line of the file.

These files are in the html_ops/ directory.