[
https://issues.apache.org/jira/browse/NUTCH-2694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16774512#comment-16774512
]
Sebastian Nagel commented on NUTCH-2694:
----------------------------------------
+1
A HostDatum has no version byte in its serialization (a CrawlDatum has), so
there is no way to do a necessary change without the need to re-create the
HostDb from the CrawDb.
Yes, a single sitemap (index) can list 2.5 billion URLs. :)
> HostDB to aggregate by long instead of integer
> ----------------------------------------------
>
> Key: NUTCH-2694
> URL: https://issues.apache.org/jira/browse/NUTCH-2694
> Project: Nutch
> Issue Type: Bug
> Components: hostdb
> Affects Versions: 1.15
> Reporter: Markus Jelsma
> Priority: Major
> Fix For: 1.16
>
>
> Last week we got Pinterest in our database, it has a neat set of sitemaps,
> and a lot of entries, over 2 billion. When first making HostDatum i foolishly
> used ints instead of longs, which shows in -1.9 billion records for Pinterest.
> I propose a simple move from int to long with an upgrade note mentioning the
> databases are not compatible and the suggestion to delete any existing
> HostDB. Agreed?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
