Thanks, now it works :) It is possible so set a default filesystem in
flink like in Hadoop (with fs.default.name)? Currently I always have to
set the complete file URI like xtreemfs://<host>:<port>/file
Best,
Lukas
Am 07.01.2015 um 12:22 schrieb Robert Metzger:
Hi Lukas,
I see that there is a XtreemFS Hadoop client
(http://www.xtreemfs.org/download.php?t=source). WIth this pending
pull request https://github.com/apache/flink/pull/268 you can use all
file systems supported by hadoop with Flink (we support the
org.apache.hadoop.FileSystems interface).
The pull request has not been merged yet because of a failing test,
but that should not affect you.
If you want, you can check out the branch of my pull request
git clone https://github.com/rmetzger/flink.git
cd flink
git checkout flink1266
mvn clean install -DskipTests
In the "flink-dist/target/flink-XXX/flink-yarn-XXX/" directory is the
finished built.
Let me know if you need more help or information.
Best,
Robert
On Wed, Jan 7, 2015 at 12:00 PM, Lukas Kairies
<[email protected] <mailto:[email protected]>>
wrote:
Hello,
I like to test flink on YARN with the alternative file system
XtreemFS. Therefore I have to add a jar file to flink but I found
no possibility to do so. How can I do this? Hadoop works fine with
XtreemFS.
Thanks
Lukas
