Hi Aaron,

All of the work done on the 186 branch was done keeping only CentOS5/RHEL5 in 
mind so as to get a stable working version. RHEL6 is something we are looking 
to test against and provide a patch for as soon as possible ( within the next 
couple of weeks ) and depending on community interest, we plan to incorporate 
support for the other flavors too.  

I started looking into RHEL6 for our internal branch copy a couple of days back 
and am attaching a quick patch if you would like to try it out ( not fully 
tested though). The patch changes os verification step and the puppet agent 
install script. The patch for now hardcodes assumptions that all nodes are 
CentOS6/RHEL6.

For the epel issue, if you face issues even after using the attached patch, one 
quick workaround option would be to remove the code that installs the 
epel-release in the puppet_agent_install script and install the correct 
epel-release manually on all nodes at the very start. 

One quick thing to note is that you should install glibc.i386 on all your nodes 
as 32-bit jdk installs fail without it ( required as datanodes/tasktrackers are 
currently configured to run in 32-bit mode ) 

The Hortonworks repo contains CentOS6/el6 packages so you should likely not 
face any issues on that front ( assuming that you have already obtained the hdp 
repo via 
http://public-repo-1.hortonworks.com/HDP-1.0.13/repos/centos6/hdp-release-1.0.13-1.el6.noarch.rpm
 )

Apart from the above, I am assuming you sorted out all the ambari rpm install 
issues and probably installed php-process for the posix_getpid errors to go 
away. Please let us know if you had to take any additional steps outside of the 
README ( a jira with a patch would be highly appreciated ).

In the meantime, please go ahead and file jiras for all the issues you come 
across with Ambari be it RHEL5 or 6. 

thanks 
-- Hitesh



On Jun 11, 2012, at 6:59 AM, Aaron Tokhy wrote:

> Hello,
> 
> I recently built Ambari (specifically the ambari-186 branch) to evaluate its 
> current state.  So far I have a web server running Ambari and I am at the 
> step where I am adding hosts to the Ambari server.  I have deployed a public 
> key on all the target instances (all of which are RHEL 6.2 LXC containers 
> with their own IP address in an internal network) and I'm trying to get it to 
> recognize them.
> 
> During "Node Discovery", it was able to perform the following:
> 
> 1) "Finding reachable nodes: All 5 nodes succeeded"
> 2) "Obtaining information about reachable nodes: All 5 nodes succeeded"
> 
> However it rejects my instances as having an unsupported OS.  The host 
> running Ambari (localhost) is CentOS 6.2 and the nodes that are a part of a 
> Hadoop/HBase/ZooKeeper cluster are all RHEL 6.2 instances.
> 
> I checked the sqlite3 database /var/db/hmc/data/data.db and found this:
> 
> sqlite> .dump Hosts
> PRAGMA foreign_keys=OFF;
> BEGIN TRANSACTION;
> CREATE TABLE "Hosts" (
>  cluster_name TEXT, -- foreign-key cluster_name
>  host_name TEXT,  -- host name
>  ip TEXT,  -- ip
>  total_mem   INTEGER,  -- total mem on host - in bytes
>  cpu_count   INTEGER,  -- cpu count - no. of cores
>  os_arch TEXT, -- os arch - i386, x86_64 etc
>  os_type TEXT, -- os type - searchable - defined set of supported OS types - 
> RHEL5, RHEL6, CENTOS5, SLES, etc.
>  os BLOB, -- os info - type, version etc - uname -a?
>  disks_info BLOB, -- disk capacity json object, keys are mount-points
>  discovery_status TEXT, -- discovery state - if we could connect properly
>  bad_health_reason TEXT, -- failure reason if any
>  rack_info TEXT, -- rack information for host
>  attributes BLOB, -- attributes blob to store additional attributes that may 
> be required for the host
>                   -- attributes stored currently:
>                      -- "publicFQDN" => public hostname in case of AWS ( same 
> as hostname in other environments )
>                      -- "privateFQDN" => private hostname in case of AWS ( 
> same as hostname in other environments )
>  PRIMARY KEY(host_name),
>  FOREIGN KEY (cluster_name) REFERENCES Clusters(cluster_name),
>  UNIQUE(ip)
> );
> INSERT INTO "Hosts" 
> VALUES('gce','slave-3','127.0.0.1',3766,2,'x86_64','redhatenterpriselinuxserver6','Red
>  Hat Enterprise Linux Server release 6.2 (Santiago);Linux rhel-4 
> 2.6.32-220.el6.x86_64 #1 SMP Wed Nov 9 08:03:13 EST 2011 x86_64 x86_64 x86_64 
> GNU/Linux','[]','FAILED','Unsupported 
> OS',NULL,'{"publicFQDN":"rhel-4","privateFQDN":"rhel-4"}');
> CREATE INDEX host_mem_index ON "Hosts" ( total_mem );
> COMMIT;
> 
> 
> How much work could it possibly be to manually (fully) add RHEL 6 support if 
> I add RHEL 6 and CentOS 6 to the supported OS types here?  I added CentOS and 
> RHEL 6 to the list of supported OS types and now it is looking to add the 
> EPEL repo for CentOS 5.
> 
> /usr/share/hmc/php/frontend/addNodes/verifyAndUpdateNodesInfo.php
> 
> Is Ambari currently only supporting the CentOS/RHEL 5/SuSE 10 series of Linux 
> distributions?
> 
> 
> Thanks!
> 
> 
> 
> 
> Aaron Tokhy | Big Data Engineer | GCE
> 
> The information contained in this e-mail and any attachment(s) is 
> Confidential, Privileged, Protected from any disclosure, and proprietary to 
> Global Computer Enterprises, Inc.  The person addressed in the email is the 
> sole authorized recipient.  If you are not the intended recipient, you are 
> hereby notified that any review, use, disclosure, retransmission, 
> dissemination, distribution, copying, or any other actions related to this 
> information is strictly prohibited. If you have received this communication 
> in error, please inform the sender and delete or destroy any copy of this 
> message.

Reply via email to