old.kete.net.nz :Documentation

For whatever reason, I get some errors stopping zebra.

kete@kohavm77:~/apps/kete_stable_dev$ sudo rake zebra:stop
(in /home/kete/apps/kete_stable_dev)
Warning: bad ps syntax, perhaps a bogus '-'? See http://procps.sf.net/faq.html

kete@kohavm77:~/apps/kete_stable_dev$ sudo rake zebra:stop
(in /home/kete/apps/kete_stable_dev)
cat: /home/kete/apps/kete_stable_dev/log/zebra.pid: No such file or directory
awk: line 1: syntax error at or near )
awk: line 1: syntax error at or near }
Warning: bad ps syntax, perhaps a bogus '-'? See http://procps.sf.net/faq.html
kill: usage: kill [-s sigspec | -n signum | -sigspec] pid | jobspec ... or kill -l [sigspec]
rm: cannot remove `/home/kete/apps/kete_stable_dev/log/zebra.pid': No such file or directory

The second error seems to indicate that the former operation succeeded. 

The errors are the same without sudo.

Tags: Zebra, Kete, zebra:start, zebra:stop, rake, ps, error

 In the file:

zebradb/zebrasrv-kill.sh

The command:

ps -o pid,ppid -ax

Might be:

 ps -o pid,ppid ax

This is on Debian etch. 

Hi Joe,

Thanks for the patch for Debian Etch.  I really appreciate when people solve their own problems and then contribute back to the software.  Great stuff.

I'll test this on Mac OS X and Solaris, if it works for all three I'll apply it as a bug fix to 1-0-stable branch.

In the future, this sort of thing probably belongs in a ticket.  You can read how to add a ticket here:

Walter 

Tags: Patches, tickets, Debian Etch

I'm looking for away to start Zebra and also the backgroundrb script automatically. Wonder if adding this to /etc/rc.local will do the trick.

/usr/local/bin/ruby /home/kete/site/script/backgroundrb start
/usr/local/bin/rake -f /home/kete/site/Rakefile /home/kete/site/ zebra:start 

Assuming your rails root is in /home/kete/site/ and your ruby is in /usr/local/ like mine is. Is there a better way anyone knows to do this?

Hi John,

It's one of those things that has been on the todo list for a long time for us. You may have beat us to the punch.

It's seems like it is in the right direction. If you could give it a try and report back, that would be great.

Tags: rake, Zebra, backgroundrb, startup

the search in non english language not accurate

 , can i use the ICU libraries with zebra and yaz search

I haven't tried.  All the zebra configuration for Kete is under zebradb/conf/.  Take a look there.  The two main configuration files that I think you will need to change are zebra-public.cfg and  zebra-private.cfg.

Let me know how you get on.  I'm curious what is necessary to get Zebra to play nicely with arabic and other languages.

hello agian,

i want to add yaz-icu functionality to the zebra

http://www.indexdata.com/yaz/doc/yaz-icu.html

to support the arabic language in zebra

it can done by adding the icu xml file the path zebradb/etc

but this bath is not found

, i asked about ,if i can use the icu with the zebra conf ??

thanks,

I would add the file to the following path and see if it kicks in:

your_app/zebradb/tab/

This should be read in as having that file in that directory take precedence because it is earlier in your profilePath as specified in your zebra-*.cfg configuration files over the the default Zebra configuration that comes with the program's configuration.

Let me know if that works or not.

Tags: Arabic Zebra, Zebra, configuration

hello ,

when i added the zebra conf file to the tab folder

this error appears

Unsupported Truncation attribute (120) 3

Tags: Zebra, icu

That actually sounds like progress. In other words we know that your icu file is taking effect.

Could you give me more context for that error please? Look for "query is" line in the log (probably log/development.log, but if you are in production mode log/production.log)  just before the error happens and paste that in here.

How long is your icu file? Is it reasonable length to paste into discussion? That would probably be helpful to track things down.

My guess is the specific issue is related to Kete's use of truncation "@attr 5=3", but I'm not positive yet. This attribute to allow for partial matches is used in a large amount of Kete queries.  In other words a search for "Bob" would match records with "Bobby" in them.

Tags: Arabic Zebra, Zebra, icu, configuration

this is the content of zebra log file :

02:52:23-29/09 zebrasrv(200) [session] Session - OK 200 tcp:::ffff:127.0.0.1 5103
02:52:23-29/09 zebrasrv(200) [request] Auth idPass kete -
02:52:23-29/09 zebrasrv(200) [request] Init OK - ID:81 Name:ZOOM-C/YAZ Version:3.0.46 0d8c31b34cd1bcf78c44ad669242f3967c76b379
02:52:23-29/09 zebrasrv(200) [request] Search public ERROR 120 1 1+0 RPN @attrset Bib-1 @attr 5=3 @attr 1=12 :Topic:
02:52:28-29/09 zebrasrv(200) [session] Connection closed by client
02:52:29-29/09 zebrasrv(201) [session] Session - OK 201 tcp:::ffff:127.0.0.1 5105
02:52:29-29/09 zebrasrv(201) [request] Auth idPass kete -
02:52:29-29/09 zebrasrv(201) [request] Init OK - ID:81 Name:ZOOM-C/YAZ Version:3.0.46 0d8c31b34cd1bcf78c44ad669242f3967c76b379
02:52:29-29/09 zebrasrv(201) [request] Search public ERROR 120 1 1+0 RPN @attrset Bib-1 @attr 5=3 @attr 1=12 :Topic:

-------------------------------------------------------------------------------------------------------------------

and yhis the guery is section in the development.log

what is query: " @attr 1=12 @attr 5=3 \":Topic:\" "
query is  @attr 1=12 @attr 5=3 ":Topic:" , syntax XML

------------------------------------------------------------------------------

and i add these files :

default.idx

contains :

index w
completeness 0
position 1
alwaysmatches 1
firstinfield 1
charmap word-phrase-utf.chr
icuchain icu.xml
#firstinfield 1

# Phrase index
# Used if completeness is 'complete {sub}field' (@attr 6=2, @attr 6=1)
# and structure is word/phrase/word-list/free-form-text/document-text
index p
completeness 1
charmap word-phrase-utf.chr
firstinfield 1
icuchain icu.xml

# URX (URL) index
# Used if structure=urx (@attr 4=104)
index u
completeness 0
charmap urx.chr
icuchain icu.xml
# Numeric index
# Used if structure=numeric (@attr 4=109)
index n
completeness 0
charmap numeric.chr
icuchain icu.xml

# Null map index (no mapping at all)
# Used if structure=key (@attr 4=3)
index 0
completeness 0
position 1
charmap @

# Year
# Used if structure=year (@attr 4=4)
index y
completeness 0
charmap @

# Date
# Used if structure=date (@attr 4=5)
index d
completeness 0
charmap @
icuchain icu.xml

# Sort register
sort s
completeness 1
charmap sort-string-utf.chr
icuchain icu.xml

---------------------------------------------------------

icu.xml  like the chain.xml in this link  http://www.indexdata.com/yaz/doc/yaz-icu.html

Tags: Zebra, Arabic, icu

Thanks for the details.

It would be good to eliminate the most likely cause first. Could you run the following from the command line from your applications root directory please?

$ rake zebra:load_initial_records # add RAILS_ENV=production if appropriate

Usually when we get an "attribute unsupported" error from Zebra it means that the indexes haven't been bootstrapped with seed records. I'm pretty sure this is not your problem, but I think it would be good to verify that that is not the cause.

hello ,

i used this command :

rake zebra:load_initial_records

but there is no change , the error is the same before

Unsupported Truncation attribute (120) 3

and the zebra log error is the same

Search public ERROR 120 1 1+0 RPN @attrset Bib-1 @attr 5=3 @attr 1=12 :Topic:

Tags: Arabic Zebra

Ok, it's good to verify that that is not the cause.

I'll see if I can replicate the issue over here. If I need further information, I 'll email you.

hello walter

i changed the truncation attr from 5=3 to 5= 100 to truncate none ,

the truncation error solved , but the search stoped completely

> hello walter
>
> i changed the truncation attr from 5=3 to 5= 100 to truncate none ,
>
> the truncation error solved , but the search stoped completely

Yep, you can turn truncation off, but that effectively breaks Kete's search functionality. It relies on truncation. I believe I have mentioned this privately in email in the past.

We have two options:

* somehow get truncation support in the ICU functionality of Zebra through that software's open source project

* fundamentally change Kete's design for search.

I'm not sure which will take more work to be honest. It is a bit of roadblock though. I would enquire with the IndexData about what it would take for them to get 5=3 truncation working with your set up.

After we get an answer on that we can evaluate which is a better answer. Unfortunately I can't currently offer resources for adding this right now. However, I am talking with a project that may be interested long term.

Wish I had better news,

Walter

hello walter ,

thanks for your interest

i want to disable the truncation , how can i do that ?

and when i use 5=100 in  QUALIFYING_ATTRIBUTE_SPECS =>  'partial' => "@attr 5=100 "

the search and indexing stops,

what can i use for disabling the truncation ?

thanks ,

Tags: icu, search

That's my point. You can't disable truncation and have Kete search work. There was a design decision (narrowing results by type such as Topic, StillImage, etc. among other things) in search that relies on truncation.

So if you disable truncation it will break Kete's searching capabilities.

See my comment above about the two options.

Tags: icu, search

Hi Walter,

I found in the web that you had encountered the captioned issue. Has this been resolved yet?

> I found in the web that you had encountered the captioned issue. Has this been
> resolved yet?

I'm afraid not. We need to work with IndexData (maintainers of Zebra) to get the ICU issue sorted. It's a significant amount of development work and we'll need to raise funds to support it.

If you would like to support this work, please contact me by email: walter -at- katipo -dot- co -dot- nz.