Problems when matching long files

56 views
Skip to first unread message

luke.e.r...@gmail.com

unread,
Apr 4, 2014, 1:39:23 PM4/4/14
to echo...@googlegroups.com
Hi everyone,

I have set up my echoprint-server and the database consists of 5 radio shows, each between an hour and two hours in duration. Whenever I query the server with a 30 second fingerprint of any of the files, starting from 0-700 seconds, I will receive a correct match. Beyond around 720 seconds, I will receive error code 7 for each query, that is the response will be that the file is not in the database. I have also reproduced this error with a database with 2 long music files.

Has anyone else come across this issue, or has an idea what could be the cause? 

Thanks
Luke

luke.e.r...@gmail.com

unread,
Apr 22, 2014, 5:19:46 PM4/22/14
to echo...@googlegroups.com
A follow up to this issue, if anyone is interested...

In testing I found that the quality of matches returned by Solr had degraded beyond around 720 seconds, to the point that the API was rejecting the results, as there were too few matching hashes in the code string.

I'm still unsure as to why this occurs for me consistently beyond 720 seconds.

To confirm, I modified the API so that the results were returned, instead of just producing the error code. They were, as expected, incorrect.   

Brian Whitman

unread,
Apr 23, 2014, 2:12:27 PM4/23/14
to echo...@googlegroups.com
It's likely depending how your solr is set up that there's a max document size and things are getting clipped on indexing. 



--
You received this message because you are subscribed to the Google Groups "echoprint" group.
To unsubscribe from this group and stop receiving emails from it, send an email to echoprint+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Brian Whitman

unread,
Apr 23, 2014, 2:16:02 PM4/23/14
to echo...@googlegroups.com
yeah, default solrconfig.xml has a maxfieldlength of 50000, which I believe is tokens (hash / time / hash / time), and you'll start seeing that get clipped after ~10m of most music. make it bigger if you really need to match super long things, or split up the thing and index separately.

B

luke.e.r...@gmail.com

unread,
Apr 23, 2014, 3:38:37 PM4/23/14
to echo...@googlegroups.com
Thanks for this Brian. I had been doing the latter, but I'll have a play with the maxfieldlength and see what happens.

Cheers,
Luke
Reply all
Reply to author
Forward
0 new messages