New product model parameters

Clustering

Guidewire has added the following new configuration parameters in PolicyCenter 9.0.5 that relate to PolicyCenter multi-node clustering:

  • ClusteringProductModelUpdateMaxRetries
  • ClusteringProductModelUpdateRetryTimeout
  • ClusteringProductModelUpdateSleepTimeCeiling

In a multi-node PolicyCenter cluster, you install product model changes on one server, then start that server to force the upgrade changes. After the first server completes the upgrade process, you can then start the other nodes in the server cluster simultaneously. Due to the number of concurrent server starts, it is possible for there to be contention for the shared update lock while updating the PolicyCenter product model on each node.

These three parameters work in conjunction with each other to manage how PolicyCenter manages lock contention in the product model upgrade process The following table illustrates the connections between these three configuration parameters.
Parameter Default Value Effect
ClusteringProductModelUpdateMaxRetries 25 retries The server attempts to acquire the database lock and fails. It then tries another n-1 times to acquire the lock until it is either successful or reaches the maximum number of permitted retries (n), at which point the lock acquisition fails.
ClusteringProductModelUpdateRetryTimeout 5 seconds The first retry attempt to acquire the lock occurs 5000 ms or 5 seconds after the first attempt. The second retry occurs 10 seconds (2 * 5) after the second attempt. The third retry occurs after 15 seconds (3 * 5). This pattern repeats until the 9th retry, after which the time between successive retries increases to 50 seconds (10 * 5).
ClusteringProductModelUpdateSleepTimeCeiling 50 seconds After the calculated time between retries equals the value of the sleep time ceiling, PolicyCenter waits this amount of time between all succeeding retry attempts until the lock process either succeeds or fails.

See the configuration parameter information in the PolicyCenter Configuration Guide for more information

Availability lookup tables

the lookup table manager uses a cache to improve performance of product model availability lookups. added a AvailabilityContextCacheEntryExpirationTime configuration parameter that sets the expiration time.

See also