bin/start-dfs.sh do not support -importCheckpoint ?

Hi all, I loo the start-dfs.sh, it seems that it only support three start mode : regular , upgrade and rollback. So why here do not support importCheckpoint ? what is the consideration ? -- Best Regards Jeff Zhang

RE: Problem compiling from source

Hi, I also have the source from hadoop-common. However, when I do ant clean jar from hdfs folder, ivy seems to try and download the hadoop-core jar file from the repository? Maybe the newer version of hadoop-core is not compatible with mine? How do I force the compile to use my own jar file, t

Re: Is it possible to recover name node just using the meta data in a new machine

Thanks Todd, I forgot to change dfs.http.address On Fri, May 21, 2010 at 12:58 PM, Todd Lipcon wrote: > On Thu, May 20, 2010 at 9:50 PM, Jeff Zhang wrote: >> >> Hi Todd, >> >> I try the command, but no process is using the port 50070 >> > > Double check fs.default.name and dfs.http.address are

Re: Is it possible to recover name node just using the meta data in a new machine

On Thu, May 20, 2010 at 9:50 PM, Jeff Zhang wrote: > Hi Todd, > > I try the command, but no process is using the port 50070 > > Double check fs.default.name and dfs.http.address are both pointing to a domain name which resolve on that machine to a local IP address. It seems to think you're trying

Re: Is it possible to recover name node just using the meta data in a new machine

Hi Todd, I try the command, but no process is using the port 50070 On Fri, May 21, 2010 at 12:39 PM, Todd Lipcon wrote: > The new machine apparently has something listening on one of the NN ports. > try sudo fuser -n tcp 50070 > (will tell you which pid is listening on that port) > -Todd > > O

Re: Is it possible to recover name node just using the meta data in a new machine

The new machine apparently has something listening on one of the NN ports. try sudo fuser -n tcp 50070 (will tell you which pid is listening on that port) -Todd On Thu, May 20, 2010 at 9:35 PM, Jeff Zhang wrote: > Hi all, > > I'd like to recover the name node in another new machine. I copy the

Is it possible to recover name node just using the meta data in a new machine

Hi all, I'd like to recover the name node in another new machine. I copy the meta data from the old name node to the new name node, and then modify the configuration (including the fs.default.name). Then I stop the old dfs cluster, and restart the new dfs cluster on the new machine, then I get the

RE: Problem compiling from source

Hi Harold, The error message "cannot find symbol" hints of not having necessary libraries. Looks like it is trying to access Kerberos libraries which it is unable to find. You can check if all the required Kerberos libraries are available. Regards, Sagar -Original Message- From:

Problem compiling from source

Hi All, For some reason, my hdfs source code can't compile anymore. ~1-2 weeks ago it was compiling fine but now it's not. I haven't made any changes to my code since I last compiled. When I do ant clean jar: I get the following errors. compile-hdfs-classes: [javac] Compiling 198 source fi

Re: Thrift gateway - status? Compared to other alternatives?

On Thu, May 20, 2010 at 09:29, Erik Forsberg wrote: > Assuming we're building a web interface that needs to read some files > from HDFS, and we don't want to use the Java API, would the Thrift > gateway be the best option? Or is fuse-dfs better? I wrote namenode and datanode plugins to expose HDF

Thrift gateway - status? Compared to other alternatives?

Hi! What's the status of the Thrift gateway for HDFS (http://wiki.apache.org/hadoop/HDFS-APIs)? My google karma seems to be bad because I'm not getting many hits from people using it. Is it stable? How's the performance? Assuming we're building a web interface that needs to read some files fro