L1 Cache in Replicated cache

[earnest.dyke]

Greetings all,

In reading the documentation I see that there is L1 (near cache) for distributed caches but I do not see that as an option for Replicated caches. Is there a reason for that? Is there a way to emulate L1 for a Replicated cache?

Thanks!

Earnie!

[nadirx]

No need: in replicated mode every node has all of the data.

[mmr11408]

Is there a required configuration parameter to enforce that in replicated mode every node has all of the data? As stated in the append below, if a replicated node keeps part of the data in memory and the rest in a file store, it is observed that depending on the nodes around it, a replicated node may get partial data. Would appreciate a response as whether that is a bug introduced in 7.x or 8.x or a configuration issue:

Two issues: All cache is not propagated properly at startup, clusters combined even though they use different UPD addresses

[dan.berindei]

Tristan is right, there is no option to control how much data each node holds in replicated mode, and there is no need for it.

However, there is a configuration parameter deciding whether state transfer should include entries in cache stores, and it's disabled by default. You need to explicitly write <file-store fetch-state="true"/> in your configuration. (The name is wrong, the state sender check this parameter, but we didn't change it because in general we're assuming all nodes use the same configuration.)

[earnest.dyke]

I think my original post was not clear. This is a matter of a remote cache that is a replicated cache. When accessing it remotely via a Hot Rod client I would like to have cache entries store locally for a short period of time thus preventing the network hop to the remote server for repeated calls for the same data.

Earnie!

[nadirx]

Ah ok, so you want to look into the Hot Rod client's Near Cache:

http://infinispan.org/docs/8.1.x/user_guide/user_guide.html#_near_caching

Tristan