Permalink
Browse files

More comprehensive log on rejected recrawls caused by date constraint

  • Loading branch information...
luccioman committed Jan 13, 2018
1 parent d47afe6 commit 3638e16c2e046d0b92feae504e8e8c1588b8cee9
Showing with 3 additions and 1 deletion.
  1. +3 −1 source/net/yacy/crawler/CrawlStacker.java
@@ -455,7 +455,9 @@ public String checkAcceptanceInitially(final DigestURL url, final CrawlProfile p
CrawlStacker.log.fine("RE-CRAWL of URL '" + urlstring + "': this url was crawled " +
((System.currentTimeMillis() - oldDate.longValue()) / 60000 / 60 / 24) + " days ago.");
} else {
return CRAWL_REJECT_REASON_DOUBLE_IN_PREFIX + ": local index, oldDate = " + ISO8601Formatter.FORMATTER.format(new Date(oldDate));
return CRAWL_REJECT_REASON_DOUBLE_IN_PREFIX + ": local index, recrawl rejected. Document date = "
+ ISO8601Formatter.FORMATTER.format(new Date(oldDate)) + " is not older than crawl profile recrawl minimum date = "
+ ISO8601Formatter.FORMATTER.format(new Date(profile.recrawlIfOlder()));
}
return null;

0 comments on commit 3638e16

Please sign in to comment.