Preserving Alert State on Restart

When the Prognosis service is restarted there is the possibility that any existing alert messages from before the restart will be repeated. With the 'Preserve Alerts' feature enabled, this situation will be prevented by storing any existing Alert information in cache so when a restart occurs it will remember what it was doing before the restart and will continue from where it left off without repeating any existing alert messages.

MS Windows and UNIX Servers

For Unix and Windows servers, persistent alert state data is stored in an 'autoancache' folder under the <Prognosis_Home>\Server\Configuration folder. A subfolder will be created for each running Threshold, with a folder name based on the name defined in the Threshold document. Separate cached data is stored for each Threshold. For Thresholds with a name containing characters that cannot be used in Windows or Unix file systems (\ / : ? " * < > |), these characters are replaced with an encoding similar to URI encoding. With this encoding, it should be simple to identify the cached data for any Threshold.

For example, a Threshold named 'IPTM:Alerts' would have its persistent alert state stored in the following folder:

..\Server\Configuration\autoancache\IPTM#3A#Alerts

HPE NonStop Servers

For HPE NonStop the data is stored in the Prognosis subvolume. The cache file name format is: 'aa' + prefix + 4 digits. For example:

'aabc0001'

Prefixes can be in the range 'aa' to '99'.

There is currently no provision to carry forward cached data when upgrading to a later version.

Enabling Preserve Alert State

The 'Preserve Alerts' feature will be enabled by default. However, if at any time it is required to be manually enabled or disabled this can be done through the ALERTS Configuration file.

Example:

SUBYSYS ALERTS
...
PRESERVE-ALERTS (Y)
CACHE-INTERVAL (30
CACHE-TIMEOUT (86400)

Clearing the Alerts Cache

The Alerts Cache can be cleared by performing any of the following actions:

  • Run 'irclean autoan' - this clears only the Alerts cache
  • Run 'irclean partial' - this will clear the Alerts cache plus any other functions of the partial clean
  • Run 'irclean full' - this will clear the Alerts cache plus any other functions of the full clean
  • Perform a 'cold' restart of the Prognosis Server
Provide feedback on this article