Document Index should be done in background

Help us to improve OpenKM! Be part of the Open Source Community.
Forum rules
Please, before asking something see the documentation wiki or use the forum search function.

Document Index should be done in background

Postby cnok » Thu Jul 08, 2010 3:29 am

Hi,

1. When I am uploading a big file. OpenKM will spend several minutes to immediately index the file. User will not be happy since they have to wait until indexing is completed so they can upload the next file(s).

OpenKM should perform document indexing in the background.

2. While exiting, OpenKM will prompt a small window for user to click 'Close'. I think this 'Close' button without any other choice is no value. I would suggest you also add 'Cancel' button as well. Otherwise, please remove this pop up window and go exit right away.

Regards,
cnok
Fresh Boarder
Fresh Boarder
 
Posts: 19
Joined: Tue Jun 08, 2010 3:03 pm

Re: Document Index should be done in background

Postby jllort » Sat Jul 10, 2010 1:25 pm

OK

I'll study ... one solution to first could be some configuration in OpenKM.cfg ... because I think this information to users could be interesting ... in your case not but some users might know that document is yet not done indexed, and information could continue for someone being interesting. To other simply disappears by OpenKM configuration.

About "Close" button ... really I could make it automatically without button ... it's really simply doing it. I'll put in our tracking system to be done.
jllort
Moderator
Moderator
 
Posts: 1828
Joined: Fri Dec 21, 2007 11:23 am
Location: Sineu - ( Illes Balears ) - Spain

Re: Document Index should be done in background

Postby cnok » Wed Jul 14, 2010 3:36 pm

In my opinion, unable to find new scanned documents during first couple minutes after they are scanned is more acceptable than keep me waiting for several minutes after each document upload. I tried uploading files with 3-5MB each, only few of them can be done in an hour. This is extremely painful. Online document system can be successful only if user who responsible for upload is convinced to do so. They should spend less time as much as possible.

Another observation, OpenKM seems to spent very long time trying to index the whole video file. Since each Video file is large, I think only properties should be extracted. In my environment, a video file of approximately 200M will need at least 5-10minutes to perform indexing. Uploading training media to OpenKM is very boring process.
cnok
Fresh Boarder
Fresh Boarder
 
Posts: 19
Joined: Tue Jun 08, 2010 3:03 pm

Re: Document Index should be done in background

Postby jllort » Wed Jul 14, 2010 10:02 pm

It's strange that 3-5MB of pdf file needs minutes ?

In other hand OpenKM in the process althought appears ( indexing ... ) really it's making something more ... for example if you've configured antivirus ... it's scanning for viruses too.

If you upload file with webdav this process is not visible as is showed in browser UI.

If you need to uploading so much files ( version 5.0 comes with uploading applet -> drag and drop folders and documents ) ... and if you can uploading zip file too and decompress it. It's other easy solution.
jllort
Moderator
Moderator
 
Posts: 1828
Joined: Fri Dec 21, 2007 11:23 am
Location: Sineu - ( Illes Balears ) - Spain

Re: Document Index should be done in background

Postby cnok » Thu Jul 15, 2010 6:15 am

Yes, it is faster if uploading via webdav. This is my favorite work around at the moment. For zip, I cannot upload zip file in okm4.1 because mine is asian language. There is a known bug to export/import zip file with non-english file name (written in a forum). I will have to wait for a new sub version or v.5 to come out.
cnok
Fresh Boarder
Fresh Boarder
 
Posts: 19
Joined: Tue Jun 08, 2010 3:03 pm

Re: Document Index should be done in background

Postby pavila » Thu Jul 15, 2010 11:11 am

Uploading big files is slow because the anti-virus verification: it have to create a file clone to check for viruses. I will try to improve the performance in the upcoming OpenKM 5 .
pavila
Moderator
Moderator
 
Posts: 836
Joined: Tue Dec 11, 2007 6:02 pm
Location: Alicante, Spain

Re: Document Index should be done in background

Postby cnok » Thu Jul 15, 2010 1:18 pm

Today I tried uploading 2 of 245MB training video to OpenKM. It took approximately 5minutes to upload and 3 minutes for indexing each of them. So, I spent approximately 16minutes total. My server is WinXP SP3 Pentium 4 2.8GHz Hyperthread 2GB RAM. Jboss and repository are in NAS which is link with the server via a gigabit network. Total repository size is 18GB (reported by OKM admin screen).

What should be optimum values of Xms, Xmx and MaxPermSize in Jboss configuration (in run.bat or run.sh and run.conf) so can obtain maximum performance from OpenKM ?

Regarding virus scan, I don't think we need to scan video files.
cnok
Fresh Boarder
Fresh Boarder
 
Posts: 19
Joined: Tue Jun 08, 2010 3:03 pm

Re: Document Index should be done in background

Postby jllort » Thu Jul 15, 2010 2:18 pm

I continue suggesting you uploading by webdav ( try this software http://www.webdrive.com/ )

In your initial post was talking about pdf files ... now are AVI with 245MB ... the scenario is changing ...

About how configuring java memory with jboss it depends about your scenario ... etc ... we only can recomend you read articles about configuring jboss memory and java memory ... it's not trivial and depends too if you've got a 32 or 64 bits OS.

I could suggest in google two queries
jbm memory
jvm memory jboss
jllort
Moderator
Moderator
 
Posts: 1828
Joined: Fri Dec 21, 2007 11:23 am
Location: Sineu - ( Illes Balears ) - Spain

Re: Document Index should be done in background

Postby cnok » Thu Jul 15, 2010 4:01 pm

I just want to give you some more example regarding file size and performance when OKM needs to spend time to do index and web UI wait for that.

I have tried webdrive but it does not work. Drive map is successful but if directory name is non english (Thai language). Web drive will not see any data in that directory. Log shows these is some internal server error and so communication is lost. For directory name in English, no problem all file in side can be listed and opened. No matter if file name is in Thai. I tried Netdrive but not work. Windows Net Use command also not work. Only Windows web folder is workable for me but as you know, it is unstable for large file.

This is another issue not related to indexing that I haven't reported to you yet. You may open another subject if want to discuss with me in more detail. I am willing to retest this again and send information to you.
cnok
Fresh Boarder
Fresh Boarder
 
Posts: 19
Joined: Tue Jun 08, 2010 3:03 pm

Re: Document Index should be done in background

Postby jllort » Fri Jul 16, 2010 4:07 pm

You must configure webdrive with UTF-8 option enabled ( by default is not enabled ). And yes, better other threat for other questions.
jllort
Moderator
Moderator
 
Posts: 1828
Joined: Fri Dec 21, 2007 11:23 am
Location: Sineu - ( Illes Balears ) - Spain

Re: Document Index should be done in background

Postby pavila » Mon Jul 19, 2010 4:02 pm

If you need a help tunning OpenKM, please contact us at http://www.openkm.com/Contact/.
pavila
Moderator
Moderator
 
Posts: 836
Joined: Tue Dec 11, 2007 6:02 pm
Location: Alicante, Spain


Return to New features discussion

Who is online

Users browsing this forum: MSN [Bot] and 1 guest