Ni Hao Ouyangshu, I suggest you to post the question to the user@ ML in order to get more chances of a helpful reply, since dev@ is where Jackrabbit developers discuss about Jackrabbit development.
HTH, zai jihan! -Simo http://people.apache.org/~simonetripodi/ http://simonetripodi.livejournal.com/ http://twitter.com/simonetripodi http://www.99soft.org/ On Thu, Dec 20, 2012 at 10:12 AM, Ouyangshu <[email protected]> wrote: > Hello, all > > > > We are research jackrabbit 2.5.2, then we find some problems on cluster need > your help. > > > > 1) Jackrabbit use lucene for metadata search, but it cannot support > distributed queries, then the cluster need synchronize the index file of > lucene to each node of cluster. The index file for each node is the whole > data, so the performance for search will more and more slowly. (we have over > 30 million records) > > 2) When we search the records which doc_state=’1’ with 1 million > records one node, the service will hang. The search sql like this: > > SELECT * > > FROM [nt:unstructured] AS s > > WHERE ISDESCENDANTNODE(s, [/root]) and doc_state='1' > > > > 3) If we remove the lucene in jackrabbit, change the key-value mode, > to use the CRUD mode and relational tables of database to replace them. Does > this method feasible? Does the architecture of jackrabbit be damaged if we > do this? > > > > Please help us to give some explanation, Thanks for your help. > > > > Ouyangshu > > Huawei Technologies Co., Ltd. > > Mobile: 18682182895 > > Email: [email protected] > > Address: Huawei Industrial Base,Bantian Longgang,ShenZhen,P.R.China > > www.huawei.com > >
