RE: Nutch not crawling all pages

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

RE: Nutch not crawling all pages

Markus Jelsma-2
Hello,

The CrawlDB does not lie, but you are two pages short of being indexed. That can happen for various different reasons and is hard to debug. But Bruno's point is valid. If you inject 50k but end up with 39k in the DB, this means some are filtered or multiple URLs were normalized back to the same.

My experience with websites generating valid URLs only, is that this assumption is almost never true. In our case, out of the thousands of sites maybe only a few of those with just a dozen URLs are free from errors, e.g. not having ambiguous URLs, redirects or 404s or otherwise bogus entries.

Markus
 
 
-----Original message-----

> From:Bruno Osiek <[hidden email]>
> Sent: Wednesday 30th October 2019 23:51
> To: [hidden email]
> Subject: Re: Nutch not crawling all pages
>
> What is the output of the inject command, ie, when you inject the 50000
> seeds justo before generating the first segment?
>
> On Wed, Oct 30, 2019 at 3:18 PM Dave Beckstrom <[hidden email]>
> wrote:
>
> > Hi Markus,
> >
> > Thank you so much for the reply and the help!  The seed URL list is
> > generated from a CMS.  I'm doubtful that many of the urls would be for
> > redirects or missing pages as the CMS only writes out the urls for valid
> > pages.  It's got me stumped!
> >
> > Here is the result of the readdb.  Not sure why the dates are wonky.  The
> > date on the server is correct.  SOLR shows 39148 pages.
> >
> > TOTAL urls:     39164
> > shortest fetch interval:        30 days, 00:00:00
> > avg fetch interval:     30 days, 00:07:10
> > longest fetch interval: 45 days, 00:00:00
> > earliest fetch time:    Mon Nov 25 07:08:00 EST 2019
> > avg of fetch times:     Wed Nov 27 18:46:00 EST 2019
> > latest fetch time:      Sat Dec 14 08:18:00 EST 2019
> > retry 0:        39164
> > score quantile 0.01:    1.8460402498021722E-4
> > score quantile 0.05:    1.8460402498021722E-4
> > score quantile 0.1:     1.8460402498021722E-4
> > score quantile 0.2:     1.8642803479451686E-4
> > score quantile 0.25:    1.8642803479451686E-4
> > score quantile 0.3:     1.960784284165129E-4
> > score quantile 0.4:     1.9663813566079454E-4
> > score quantile 0.5:     2.0251113164704293E-4
> > score quantile 0.6:     2.037905069300905E-4
> > score quantile 0.7:     2.1473052038345486E-4
> > score quantile 0.75:    2.1473052038345486E-4
> > score quantile 0.8:     2.172968233935535E-4
> > score quantile 0.9:     2.429802336152917E-4
> > score quantile 0.95:    2.4354603374376893E-4
> > score quantile 0.99:    2.542474209925616E-4
> > min score:      3.0443254217971116E-5
> > avg score:      7.001118352666182E-4
> > max score:      1.3120110034942627
> > status 2 (db_fetched):  39150
> > status 3 (db_gone):     13
> > status 4 (db_redir_temp):       1
> > CrawlDb statistics: done
> >
> >
> >
> > On Wed, Oct 30, 2019 at 4:01 PM Markus Jelsma <[hidden email]>
> > wrote:
> >
> > > Hello Dave,
> > >
> > > First you should check the CrawlDB using readdb -stats. My bet is that
> > > your set contains some redirects and gone (404), or transient errors. The
> > > number for fetched and notModified added up should be about the same as
> > the
> > > number of documents indexed.
> > >
> > > Regards,
> > > Markus
> > >
> > >
> > >
> > > -----Original message-----
> > > > From:Dave Beckstrom <[hidden email]>
> > > > Sent: Wednesday 30th October 2019 20:00
> > > > To: [hidden email]
> > > > Subject: Nutch not crawling all pages
> > > >
> > > > Hi Everyone,
> > > >
> > > > I googled and researched and I am not finding any solutions.  I'm
> > hoping
> > > > someone here can help.
> > > >
> > > > I have txt files with about 50,000 seed urls that are fed to Nutch for
> > > > crawling and then indexing in SOLR.  However, it will not index more
> > than
> > > > about 39,000 pages no matter what I do.   The robots.txt file gives
> > Nutch
> > > > access to the entire site.
> > > >
> > > > This is a snippet of the last Nutch run:
> > > >
> > > > nerator: starting at 2019-10-30 14:44:38
> > > > Generator: Selecting best-scoring urls due for fetch.
> > > > Generator: filtering: false
> > > > Generator: normalizing: true
> > > > Generator: topN: 80000
> > > > Generator: 0 records selected for fetching, exiting ...
> > > > Generate returned 1 (no new segments created)
> > > > Escaping loop: no more URLs to fetch now
> > > >
> > > > I ran that crawl about 5 or 6  times.  It seems to index about 6,000
> > > pages
> > > > per run.  I planned to keep running it until it hit the 50,000+ page
> > mark
> > > > which would indicate that all of the pages where indexed.  That last
> > run
> > > it
> > > > just ended without crawling anything more.
> > > >
> > > > Below are some of the potentially relevent config settings.  I removed
> > > the
> > > > "description" for brevity.
> > > >
> > > > <property>
> > > >   <name>http.content.limit</name>
> > > >   <value>-1</value>
> > > > </property>
> > > > <property>
> > > >  <name>db.ignore.external.links</name>
> > > >  <value>true</value>
> > > > </property>
> > > > <property>
> > > >  <name>db.ignore.external.links.mode</name>
> > > >  <value>byDomain</value>
> > > > </property>
> > > > <property>
> > > >   <name>db.ignore.internal.links</name>
> > > >   <value>false</value>
> > > > </property>
> > > > <property>
> > > >   <name>db.update.additions.allowed</name>
> > > >   <value>true</value>
> > > >  </property>
> > > >  <property>
> > > >  <name>db.max.outlinks.per.page</name>
> > > >   <value>-1</value>
> > > >  </property>
> > > >  <property>
> > > >   <name>db.injector.overwrite</name>
> > > >   <value>true</value>
> > > >  </property>
> > > >
> > > > Anyone have any suggestions?  Its odd that when you give nutch a
> > specific
> > > > list of urls to be crawled that it wouldn't crawl all of them.
> > > >
> > > > I appreicate any help you can offer.   Thank you!
> > > >
> > > > --
> > > > *Fig Leaf Software is now Collective FLS, Inc.*
> > > > *
> > > > *
> > > > *Collective FLS, Inc.*
> > > >
> > > > https://www.collectivefls.com/ <https://www.collectivefls.com/>
> > > >
> > > >
> > > >
> > > >
> > >
> >
> > --
> > *Fig Leaf Software is now Collective FLS, Inc.*
> > *
> > *
> > *Collective FLS, Inc.*
> >
> > https://www.collectivefls.com/ <https://www.collectivefls.com/>
> >
> >
> >
> > --
> Enviado de dispositivo móvel.
>
Reply | Threaded
Open this post in threaded view
|

Re: Nutch not crawling all pages

Dave Beckstrom-2
You guys were right.  We have one seed URL file which lists urls to 10
pages,  Each of those 10 pages has roughly 5,000 urls to be crawled.

The links to 3 out of the 10 pages were wrong (missing) -- which accounts
for roughly 15,000+ urls that were missing.  I didn't catch it because
there are multiple servers involved and everything was correct on the
server I was working on but on one of the other servers they were wrong.

I didn't set that part up so that was a mistake someone made before my
time.  But you guys clued me in to it.

Thank you!




On Wed, Oct 30, 2019 at 6:11 PM Markus Jelsma <[hidden email]>
wrote:

> Hello,
>
> The CrawlDB does not lie, but you are two pages short of being indexed.
> That can happen for various different reasons and is hard to debug. But
> Bruno's point is valid. If you inject 50k but end up with 39k in the DB,
> this means some are filtered or multiple URLs were normalized back to the
> same.
>
> My experience with websites generating valid URLs only, is that this
> assumption is almost never true. In our case, out of the thousands of sites
> maybe only a few of those with just a dozen URLs are free from errors, e.g.
> not having ambiguous URLs, redirects or 404s or otherwise bogus entries.
>
> Markus
>
>
> -----Original message-----
> > From:Bruno Osiek <[hidden email]>
> > Sent: Wednesday 30th October 2019 23:51
> > To: [hidden email]
> > Subject: Re: Nutch not crawling all pages
> >
> > What is the output of the inject command, ie, when you inject the 50000
> > seeds justo before generating the first segment?
> >
> > On Wed, Oct 30, 2019 at 3:18 PM Dave Beckstrom <
> [hidden email]>
> > wrote:
> >
> > > Hi Markus,
> > >
> > > Thank you so much for the reply and the help!  The seed URL list is
> > > generated from a CMS.  I'm doubtful that many of the urls would be for
> > > redirects or missing pages as the CMS only writes out the urls for
> valid
> > > pages.  It's got me stumped!
> > >
> > > Here is the result of the readdb.  Not sure why the dates are wonky.
> The
> > > date on the server is correct.  SOLR shows 39148 pages.
> > >
> > > TOTAL urls:     39164
> > > shortest fetch interval:        30 days, 00:00:00
> > > avg fetch interval:     30 days, 00:07:10
> > > longest fetch interval: 45 days, 00:00:00
> > > earliest fetch time:    Mon Nov 25 07:08:00 EST 2019
> > > avg of fetch times:     Wed Nov 27 18:46:00 EST 2019
> > > latest fetch time:      Sat Dec 14 08:18:00 EST 2019
> > > retry 0:        39164
> > > score quantile 0.01:    1.8460402498021722E-4
> > > score quantile 0.05:    1.8460402498021722E-4
> > > score quantile 0.1:     1.8460402498021722E-4
> > > score quantile 0.2:     1.8642803479451686E-4
> > > score quantile 0.25:    1.8642803479451686E-4
> > > score quantile 0.3:     1.960784284165129E-4
> > > score quantile 0.4:     1.9663813566079454E-4
> > > score quantile 0.5:     2.0251113164704293E-4
> > > score quantile 0.6:     2.037905069300905E-4
> > > score quantile 0.7:     2.1473052038345486E-4
> > > score quantile 0.75:    2.1473052038345486E-4
> > > score quantile 0.8:     2.172968233935535E-4
> > > score quantile 0.9:     2.429802336152917E-4
> > > score quantile 0.95:    2.4354603374376893E-4
> > > score quantile 0.99:    2.542474209925616E-4
> > > min score:      3.0443254217971116E-5
> > > avg score:      7.001118352666182E-4
> > > max score:      1.3120110034942627
> > > status 2 (db_fetched):  39150
> > > status 3 (db_gone):     13
> > > status 4 (db_redir_temp):       1
> > > CrawlDb statistics: done
> > >
> > >
> > >
> > > On Wed, Oct 30, 2019 at 4:01 PM Markus Jelsma <
> [hidden email]>
> > > wrote:
> > >
> > > > Hello Dave,
> > > >
> > > > First you should check the CrawlDB using readdb -stats. My bet is
> that
> > > > your set contains some redirects and gone (404), or transient
> errors. The
> > > > number for fetched and notModified added up should be about the same
> as
> > > the
> > > > number of documents indexed.
> > > >
> > > > Regards,
> > > > Markus
> > > >
> > > >
> > > >
> > > > -----Original message-----
> > > > > From:Dave Beckstrom <[hidden email]>
> > > > > Sent: Wednesday 30th October 2019 20:00
> > > > > To: [hidden email]
> > > > > Subject: Nutch not crawling all pages
> > > > >
> > > > > Hi Everyone,
> > > > >
> > > > > I googled and researched and I am not finding any solutions.  I'm
> > > hoping
> > > > > someone here can help.
> > > > >
> > > > > I have txt files with about 50,000 seed urls that are fed to Nutch
> for
> > > > > crawling and then indexing in SOLR.  However, it will not index
> more
> > > than
> > > > > about 39,000 pages no matter what I do.   The robots.txt file gives
> > > Nutch
> > > > > access to the entire site.
> > > > >
> > > > > This is a snippet of the last Nutch run:
> > > > >
> > > > > nerator: starting at 2019-10-30 14:44:38
> > > > > Generator: Selecting best-scoring urls due for fetch.
> > > > > Generator: filtering: false
> > > > > Generator: normalizing: true
> > > > > Generator: topN: 80000
> > > > > Generator: 0 records selected for fetching, exiting ...
> > > > > Generate returned 1 (no new segments created)
> > > > > Escaping loop: no more URLs to fetch now
> > > > >
> > > > > I ran that crawl about 5 or 6  times.  It seems to index about
> 6,000
> > > > pages
> > > > > per run.  I planned to keep running it until it hit the 50,000+
> page
> > > mark
> > > > > which would indicate that all of the pages where indexed.  That
> last
> > > run
> > > > it
> > > > > just ended without crawling anything more.
> > > > >
> > > > > Below are some of the potentially relevent config settings.  I
> removed
> > > > the
> > > > > "description" for brevity.
> > > > >
> > > > > <property>
> > > > >   <name>http.content.limit</name>
> > > > >   <value>-1</value>
> > > > > </property>
> > > > > <property>
> > > > >  <name>db.ignore.external.links</name>
> > > > >  <value>true</value>
> > > > > </property>
> > > > > <property>
> > > > >  <name>db.ignore.external.links.mode</name>
> > > > >  <value>byDomain</value>
> > > > > </property>
> > > > > <property>
> > > > >   <name>db.ignore.internal.links</name>
> > > > >   <value>false</value>
> > > > > </property>
> > > > > <property>
> > > > >   <name>db.update.additions.allowed</name>
> > > > >   <value>true</value>
> > > > >  </property>
> > > > >  <property>
> > > > >  <name>db.max.outlinks.per.page</name>
> > > > >   <value>-1</value>
> > > > >  </property>
> > > > >  <property>
> > > > >   <name>db.injector.overwrite</name>
> > > > >   <value>true</value>
> > > > >  </property>
> > > > >
> > > > > Anyone have any suggestions?  Its odd that when you give nutch a
> > > specific
> > > > > list of urls to be crawled that it wouldn't crawl all of them.
> > > > >
> > > > > I appreicate any help you can offer.   Thank you!
> > > > >
> > > > > --
> > > > > *Fig Leaf Software is now Collective FLS, Inc.*
> > > > > *
> > > > > *
> > > > > *Collective FLS, Inc.*
> > > > >
> > > > > https://www.collectivefls.com/ <https://www.collectivefls.com/>
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > >
> > > --
> > > *Fig Leaf Software is now Collective FLS, Inc.*
> > > *
> > > *
> > > *Collective FLS, Inc.*
> > >
> > > https://www.collectivefls.com/ <https://www.collectivefls.com/>
> > >
> > >
> > >
> > > --
> > Enviado de dispositivo móvel.
> >
>

--
*Fig Leaf Software is now Collective FLS, Inc.*
*
*
*Collective FLS, Inc.* 

https://www.collectivefls.com/ <https://www.collectivefls.com/