On 6/3/07, Emmanuel JOKE <[EMAIL PROTECTED]> wrote:
yes i did.
<property>
<name>io.seqfile.compression.type</name>
<value>BLOCK</value>
</property>
That's strange. I may be wrong but I don't think this is a native-lib
problem. Even if you don't have native libs in place, hadoop will fall
back to java code(at least for gzip) and do compression. It will just
be very slow:).
Can you create an OutputFormat in CrawlDbMerger and set compression
type to BLOCK manually? You can take a look at ParseOutputFormat's
code as an example.
Any clues ?
--
Doğacan Güney
