purepennons / gss

Automatically exported from code.google.com/p/gss
Other
0 stars 0 forks source link

Solr 1.3.0 usage - search #18

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago
What steps will reproduce the problem?
1. Copy solr directory from example directory to jboss server
2. Start jboss server
3. Try searching for files

What is the expected output? What do you see instead?

Expected output is Internal server error message.

What version of the product are you using? On what operating system?

OS is Centos 5 x86_64, gss is recent source, other requirements are versions 
that are listed in the wiki page for building and deploying. Solr is version 
1.3.0, Jboss is version 5.1.0 GA with hornetQ

When I run solr from example directory as standalone, running on port 8983, 
then search seems to work (no error message is displayed), but also no results 
are displayed.

Also, I had error when trying to patch solr according to wiki page, so I add 
this manually to solrconfig.xml file after /update/csv

<!-- Rich document update handler, loaded on demand -->
   <requestHandler name="/update/rich" class="solr.RichDocumentRequestHandler" startup="lazy" />

Original issue reported on code.google.com by ngara...@gmail.com on 2 Jul 2010 at 11:26

GoogleCodeExporter commented 8 years ago
I found solr lines in gss.properties file, will try to edit them to make solr 
work under jboss (port 8080).

Original comment by ngara...@gmail.com on 5 Jul 2010 at 10:09

GoogleCodeExporter commented 8 years ago
Yes, you can configure the server to connect to the solr instance you want by 
tweaking the gss.properties entries. The requestHandler entry you mentioned 
above is also correct, not sure why the patching failed.

We're going to migrate to Solr 1.4 at some point in the future and use its 
internal rich document indexing, avoiding the need to patch solr.

Original comment by past...@gmail.com on 5 Jul 2010 at 11:16

GoogleCodeExporter commented 8 years ago
Here is part of log file, uploading a file and trying to search for it. Search 
returns nothing. Am I missing something or what?

Original comment by ngara...@gmail.com on 5 Jul 2010 at 1:16

Attachments:

GoogleCodeExporter commented 8 years ago
I can see the requests to search for "zip" and its variants and the fact that 
solr returns empty-handed, but I can't find the actual upload request in the 
log. There should have been a few lines above these that contain a POST or PUT 
request and the subsequent interaction with solr.

Original comment by past...@gmail.com on 5 Jul 2010 at 1:59

GoogleCodeExporter commented 8 years ago
So, here should be POST included.
I dont see interaction with solr after POST. At least, nothing with .solr. is 
mentioned.
Also, I see some error regarding gss-indexingQueue. That queue is used by 
hornetQ?

Original comment by ngara...@gmail.com on 6 Jul 2010 at 7:44

Attachments:

GoogleCodeExporter commented 8 years ago
The gss-indexingQueue is your problem. This is the queue used by JMS (HornetQ) 
to push the indexing tasks to be performed after the user request returns. The 
queue handler will send requests to Solr to index the files.  If you have used 
"ant install" to set up the initial jboss configuration it should have copied 
jboss/deploy/hornetq.sar/hornetq-jms.xml into the appropriate jboss deployment 
directory (jboss/server/default/deploy/hornetq.sar/hornetq-jms.xml). Try to run 
"ant install" again and see if this fixes it.

You may also verify that the indexing queue is set up properly, by visiting the 
JMX console in http://127.0.0.1:8080/jmx-console/ and clicking org.hornetq in 
the bottom left.

Original comment by past...@gmail.com on 6 Jul 2010 at 10:07

GoogleCodeExporter commented 8 years ago
I am actually running jboss server (default-with-hornetq) since I think default 
is not using hornetQ, at least looking at hornetQ connection refused errors in 
log file.

So, I copied hornet-jms.xml file from default jboss instance (it has gss queue) 
to default-with-hornetQ instance, and search now seems to work. I can get 
search results based on query. Only question is why search returns no output if 
I enter few letters from the middle of the filename.

For example, filename is: Best_cyclist.wmv
And if I enter: best i get result, if I enter wmv I get result, but anything 
else like *est_cyclist* or like it returns nothing. Is that ok?

Also, I just found out, if I put "." (dot without double quotes) inside search, 
I get message that my session is expired.

Thanks in advance.

Original comment by ngara...@gmail.com on 6 Jul 2010 at 10:55

GoogleCodeExporter commented 8 years ago
Ah, OK. Running "ant install" makes the default configuration pretty similar to 
default-with-hornetq, so if you are fine with the latter, don't bother. 

When indexing files, the system tokenizes the file name, tags, contents, etc. 
and then matches the search string against these tokens, when searching. So, 
"Best_cyclist.wmv" would be tokenized as "best, cyclist, wmv, 
best_cyclist.wmv". Partial matches are not supported, nor are wildcards, so 
"*est_cyclist*" is not expected to match. It works pretty much like Google's 
search.

The "." query is obviously handled wrong. Open a new issue on it if you don't 
mind and we'll fix it.

Original comment by past...@gmail.com on 6 Jul 2010 at 11:10

GoogleCodeExporter commented 8 years ago
Ok, thank you very much for help and support with solr and clarification on 
search(ing)!

Original comment by ngara...@gmail.com on 6 Jul 2010 at 11:17