All the legacy caches listed in kRunCacheManager.sh but one (Scoring Criteria) have been converted to new caches.

All these caches are now configured from cache.conf, and their parameters in tune.conf (or server.conf/Administration Console) have been deprecated and are not used anymore.

Some parameters in cache.conf have been removed in an attempt to simplify cache configuration.

• The only parameter that can be tuned for all caches is the cache size (the parameter “maxSize”).
• Parameters “maxAge”, “maxIdleTime”, “resolver” and “stalenessCheckGraceInterval” have been removed.
• Parameter “distributed” is inferred automatically based on whether a staleness check is defined (distributed = false) or not (distributed = true).
• Cache is automatically set to disabled if maxSize = 0.

A staleness check has also been added for request types and table components even though it is not enabled by default. If you rely on direct DB updates to modify request types or table components, it is preferable to enable the staleness check rather than disable the cache. Staleness check will only work if the column LAST_UPDATE_DATE is modified during data update.

One side effect of cache simplification is that the cache maintenance thread (previously used to enforce maxAge and maxIdleTime of cached objects) is now only used to reload the cache.conf configuration when modified at runtime. It does not cause issues anymore when running at the same time as kRunCacheManager.sh.

• Flush counts of types “old” and “idle” have been removed.

• “Cache Flush All” count has been added. It displays the number of times of the whole cache being flushed. It helps identify abuses of “kRunCacheManager.sh A” or of unnecessary full caches flushes.

  Running “kRunCacheManager.sh A” will increment this value by 1 for all caches.

• The order of listed caches has changed: it still lists legacy cache followed by new caches, but now the new caches are ordered alphabetically by cache name.
• When all caches are listed, the cache names of new caches are displayed in parenthesis after the cache title.
• When flushing new caches, you can now use the cache name in place of the cache number. It is advised to always use the cache name, as it does not rely on any ordering of caches in that list.

• An extra optional parameter has been added to the command, to pass the entity ID to flush. So you can now flush one specific entity from a cache using the following syntax:

  sh ./kRunCacheManager.sh [<URL>] <cache name (or cache number)> [<entity ID>]

• A new action (E) has been added. It lists all keys in each of the new caches along with their type (String or Integer). This can help diagnose whether a specific entity is currently stored in the cache or not. Note that in MLU environments, the same key can be displayed multiple times if it is stored in the cache using different languages.

• Passing an entity ID to ksc_flush_cache will now work regardless of the key type (String or Integer).

• You can now flush the legacy caches by passing their cache number (as in kRunCacheManager.sh) instead of the cache name. Since there is only one legacy cache left in 9.31, it means passing “1” as the cache name flushes the “Scoring Criteria” cache.