nutch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roland (JIRA)" <>
Subject [jira] [Commented] (NUTCH-1511) Metadata in MYSQL updated with 'garbage'
Date Sat, 09 Feb 2013 14:59:13 GMT


Roland commented on NUTCH-1511:

Hi Lewis,

I'm not absolutly sure what you mean.
Actually I'm running a crawl job for 400k pages, so I can't reproduce this, but I can try
to describe:
Now running without scoring-opic plugin I get this from cassandra:
[default@webpage] get sc['de.welt.www:http/wams_print/article1466267/Agravis-sucht-Investoren.html'][mtdt];
=> (column=_csh_, value=<f�, timestamp=1360365320407004)
=> (column=author, value=Guido Hartmann , timestamp=1360365320408000)
=> (column=date, value=2007-12-16T04:00:00+01:00        , timestamp=1360365320408004)
=> (column=description, value=Der Handelskonzern aus Münster setzt im Jahr mit landwirtschaftlichen
Produkten und Landtechnik knapp vier Milliarden Euro um. Doch die Gewinne sind bislang mager.
Gleichzeitig steigen die Risiken im    , timestamp=1360365320409000)
=> (column=last-modified, value=2011-11-16T07:35:26+01:00       , timestamp=1360365320408002)
=> (column=location,      , timestamp=1360365320407000)
=> (column=robots, value=index,follow,noodp     , timestamp=1360365320407002)
from an old run (nutch with scorint-opic + hbase) I have this data:
hbase(main):006:0> get 'webpage', 'de.welt.www:http/wams_print/article1466267/Agravis-sucht-Investoren.html',
COLUMN                          CELL
 mtdt:_csh_                     timestamp=1360335641708, value=1U#L

I can do another exact problem report with cassandra later.

> Metadata in MYSQL updated with 'garbage'
> ----------------------------------------
>                 Key: NUTCH-1511
>                 URL:
>             Project: Nutch
>          Issue Type: Bug
>          Components: fetcher, injector, storage
>    Affects Versions: 2.1
>         Environment: Ubuntu 12.04
>            Reporter: J. Gobel
>              Labels: metadata, mysql, nutch, scoring-opic
>             Fix For: 2.2
> After applying patch for Metadata parser (NUTCH-1478) I notice that the metadata field
just before the crawl ends is populated with the correct information. However when the crawl
is completely finished the metadata field is populated with 'garbage' _csh_�����

> I notice in my SQL log file that the scoring plugin is overwriting the metadata field
in a final data insertion with '_csh_ \0\0\0\0\'. When I remove 'scoring-opic' out of 'plugin.includes'
property in the nutch-site.xml , the metadata-field is crisp and clear.
> MYSQL LOG FILE: (I did a crawl on Below you will see a fragments
of my MYSQL log file, only the moments when data is written to the METADATA field in the MYSQL
> First Insertion .. here I suppose scoring-opic writes its information, _csh_ ?€\0\0\0

> 58 Query    INSERT INTO webpage (fetchInterval,fetchTime,id,markers,metadata,score )VALUES
(2592000,1357122976493,'org.apache.nutch:http/',' dist 0 _injmrk_ y\0','
> _csh_ ?€\0\0\0',1.0) ON DUPLICATE KEY UPDATE fetchInterval=2592000,fetchTime=1357122976493,markers='
dist 0 _injmrk_ y\0',metadata='
> _csh_ ?€\0\0\0',score=1.0
> Second Insertion - inhere scraped metada is inserted into metadata. 
>  81 Query    INSERT INTO webpage (id,markers,metadata,outlinks,parseStatus,signature,text,title
)VALUES ('org.apache.nutch:http/',
> The final insertion -  please note that here the metadata field is overwritten with _CSH_\0\0\0\0
> 90 Query    INSERT INTO webpage (fetchTime,id,inlinks,markers,metadata )VALUES (1359714995075,'org.apache.nutch:http/','
> Nutch\0',' dist 0 _injmrk_ y _updmrk_*1357122982-1745626508 __prsmrk__*1357122982-1745626508
_gnmrk_*1357122982-1745626508 _ftcmrk_*1357122982-1745626508\0','
> _csh_ \0\0\0\0\0') ON DUPLICATE KEY UPDATE fetchTime=1359714995075,inlinks=' 0

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message