Arber, See my patch in https://issues.apache.org/jira/browse/HBASE-1279
Now setting hbase.regionserver should work. J-D On Fri, Mar 20, 2009 at 8:44 AM, Jean-Daniel Cryans <[email protected]> wrote: > I'm currently looking at the code and I think I see something that may > be fixing that problem. We have a similar problem here so I'll check > if it fixes it. > > J-D > > On Fri, Mar 20, 2009 at 5:51 AM, Yabo-Arber Xu <[email protected]> > wrote: >> Hi J-D, >> >> I further found that the region server can actually be connected via Web UI >> from an client outside Amazon network. That further verify the view that >> Master/Region server both are working, and it's just that Master mapped >> region servers' external IP into internal IP automatically, and the clients >> fail to communicate with region server via the mapped internal IP. >> >> Do you see any possible solution for this in the near future? >> >> Best, >> Arber >> >> On Thu, Mar 19, 2009 at 4:48 PM, Yabo-Arber Xu >> <[email protected]>wrote: >> >>> Thx for your explanation. I suspect the reason is that when the master >>> initiate the contact with region server, region server was recognized >>> through its internal address ( as both of them are on Amazon network), even >>> i explicitly put the external address for region server. >>> >>> So the consequence is is that the internal address shows in namenode, to >>> which external clients can not connect. >>> >>> >>> On Thu, Mar 19, 2009 at 4:33 PM, Jean-Daniel Cryans >>> <[email protected]>wrote: >>> >>>> Getting the "good" host name for a node is kind of a pain. For the >>>> moment, the implemented solution is that the Master tells the region >>>> server to override it's known address with what the region server was >>>> able to contact the master with. It was implemented like this so that >>>> the region servers stop showing all up as "127.0.0.1" and hadoop was >>>> doing it the same way. >>>> >>>> Now, I know that they changed it on their side, so I'd really really >>>> like to know with which address the datanodes are showing up in the >>>> namenode. Internal or external? If it's the external one, I'll dive in >>>> their code. >>>> >>>> Thx, >>>> >>>> J-D >>>> >>>> On Thu, Mar 19, 2009 at 7:26 PM, Yabo-Arber Xu <[email protected]> >>>> wrote: >>>> > Hi J-D, >>>> > >>>> > Thanks for your reply, and sorry for my late response as I was engaged >>>> in >>>> > sth else in the past two days. >>>> > >>>> > Yes, I've unlocked the port, and i am actually able to access from the >>>> web >>>> > UI with a client not running on EC2 to HBase at example.com:60010. It >>>> shows >>>> > all User Tables, but the Region Servers Address is the EC2 internal >>>> address: >>>> > domU-12-31-39-00-65-E5.compute-1.internal:60020. >>>> > >>>> > I guess the client fails because it can not connect region server, which >>>> > serves only for an internal IP. However, in hbase-site.xml, I did >>>> configure >>>> > with region server explicitly in its external IP. >>>> > >>>> > <property>^M >>>> > <name>hbase.regionserver</name> >>>> > <value>ec2-67-202-57-127.compute-1.amazonaws.com:60020</value> >>>> > <description>The host and port a HBase region server runs at.^M >>>> > </description> >>>> > </property> >>>> > >>>> > What could I do wrong? >>>> > >>>> > Thanks again, >>>> > Arber >>>> > >>>> > >>>> > On Tue, Mar 17, 2009 at 5:05 AM, Jean-Daniel Cryans < >>>> [email protected]>wrote: >>>> > >>>> >> Arber, >>>> >> >>>> >> There are security policies with EC2, did you unblock port 60000 for >>>> >> your own IP address? >>>> >> >>>> >> J-D >>>> >> >>>> >> On Tue, Mar 17, 2009 at 4:07 AM, Yabo-Arber Xu < >>>> [email protected]> >>>> >> wrote: >>>> >> > Hi all, >>>> >> > >>>> >> > I set up a small HBase cluster on EC2. It works fine internally if >>>> all >>>> >> the >>>> >> > applications are within EC2. It, however, does not work if i am >>>> trying >>>> >> run >>>> >> > shell or client program on a external host. I've been googling for a >>>> >> while, >>>> >> > and found there was similar issues raised before: >>>> >> > >>>> >> >>>> http://www.nabble.com/Hbase-on-EC2-and-issues-with-Amazon-NAT-Internal-Addresses-td21621367.html >>>> >> , >>>> >> > >>>> >> > >>>> >> > But it seems no solutions so far. I wonder whether anybody has made >>>> >> progress >>>> >> > on this issue. Any thoughts would be greatly appreciated! >>>> >> > >>>> >> > >>>> >> > Attached with my hbase-site.xml. I am running the master/region >>>> server >>>> >> all >>>> >> > on one instance for testing. >>>> >> > >>>> >> > <configuration> >>>> >> > <property> >>>> >> > <name>hbase.master</name> >>>> >> > <value>*domU-12-31-39-00-E0-96.compute-1.internal*:60000</value> >>>> //It >>>> >> > does not work even if i changed this the Amazon public IP >>>> >> > <description>The host and port that the HBase master runs at. >>>> >> > </description> >>>> >> > </property> >>>> >> > >>>> >> > <property> >>>> >> > <name>hbase.rootdir</name> >>>> >> > >>>> >> > >>>> >> >>>> <value>hdfs://domU-12-31-39-00-E0-96.compute-1.internal:54310/hbase</value> >>>> >> > <description>The directory shared by region servers. >>>> >> > </description> >>>> >> > </property> >>>> >> > </configuration> >>>> >> > >>>> >> > >>>> >> > Thanks for your attention, >>>> >> > Arber >>>> >> > >>>> >> >>>> > >>>> >>> >> >
