topic search returns different number of files every time. in Alfresco Archive

search returns different number of files every time.

supratim — Fri, 22 Jan 2010 05:41:14 GMT

Hi all,We have a webscript which extends BWAbstractRepositoryWebScript class.We have a search code which basically searches all documents withcertain extensions i.e. fileType="*.jpg" and so on protected List<FileInfo> getList(String path, String fileType,

Re: search returns different number of files every time.

dhalupa — Fri, 22 Jan 2010 07:40:04 GMT

This discussion might help you. Take a look at the last post

Kind regards,

Denis

Re: search returns different number of files every time.

supratim — Fri, 22 Jan 2010 16:27:17 GMT

Thank you Denis,
I have joined the linkedin group, my membership is subjected to approval of group manager!
Can you please point me to some other resource for the time being?
Many thanks, I really appreciate your help.

Re: search returns different number of files every time.

dhalupa — Fri, 22 Jan 2010 17:38:25 GMT

Ok, copy/paste…

I was also intrigued by this behaviour, and have looked into the code a bit. Here are my conclusions..
There are few reasons for such behaviour. When search is initated using SearchService, first thing obviously is to query lucene index with passed query. This will return a collection of lucene search hits. A search hit contains node id which satisfies search expression. This is pretty quick process but this collection knows nothing about user rights on particular nodes. Therefore acl evaluation on each hit has to be performed. There are few parameters which affect this process

(1) system.acl.maxPermissionCheckTimeMillis defaults to 10000ms and defines maximum time that will be spent on ACL evaluation
(2) system.acl.maxPermissionChecks defaults to 1000 and defines maximum number of ACL evaluations that will be performed for collection of hits that will be returned from Lucene query. If the number of hits exceedes this number, the rest will simply be discarded
(3) number of ACLs held in cache which defaults to 1000

ACL evaluation process will go through all hits from collection, if ACL is present in the cache, than cached ACL will be used. If not, ACL will be retreived from database. This means that each ACL which is not present in the cache will result in a database query. Before evaluation of each hit, it will be checked whether maximum time(1) or maximum number(2) has been exceeded. If so, evaluation will be stopped and rest of hits will be discarded.

Looking into above process, the reasons for behaviour mentioned in previous post can be understood. It is also clear that this behaviour can be affected by changing the values of those three parameteres.

Kind regards,

Denis

Re: search returns different number of files every time.

supratim — Fri, 22 Jan 2010 20:14:27 GMT

wow! That worked! Thanks a lot. Alfresco rocks, so do you! Cheers!

Re: search returns different number of files every time.

gilles — Fri, 16 Apr 2010 13:29:18 GMT

Hi, what do you mean by "That worked"? Did you manage to fetch all the nodes?
What config did you use?
What config could I use for a 5000 folders + 5000 contents repository? Any benchmarks?
If I "lucene" search in the node browser I often get exactly 101 results but sometimes more. Is there a limit behind this?
What's the better way to be sure to fecth all matching nodes?
Thanks in advance,
Gilles
(Alfresco 2.9)