phoenix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steve Terrell <sterr...@oculus360.us>
Subject Re: Phoenix 4.6.0: sqlline.py Hangs From Remote Host
Date Tue, 08 Dec 2015 00:16:53 GMT
I may have figured it out.  I found out that when I created another EMR
cluster exactly the same as my working one (same AMI, Hbase version,
Phoenix version), the problem was duplicated.  I could still connect to the
first cluster's Phoenix end point, but got timeout on the second cluster.

So what was different between the two clusters?

For my cluster that has no sqlline issues, the private ip addresses for my
master and region servers were in my local /etc/hosts file. Looked
something like this:
10.0.100.38    ip-10-0-100-38.ec2.internal
10.0.100.54    ip-10-0-100-54.ec2.internal
10.0.100.55    ip-10-0-100-55.ec2.internal
10.0.100.56    ip-10-0-100-56.ec2.internal

Once I added the ips for  servers on the new cluster, it worked.  It will
be some time before I have a chance to see if this solves my original
problem with the newer Phoenix and HBase.  but wanted to share what I
learned in case someone else is scratching their head.

Meanwhile, does anyone know why the region server ips are important?  I
thought communication was only between the client and the master node.

Thanks,
    Steve

On Sun, Nov 1, 2015 at 9:29 AM, Steve Terrell <sterrell@oculus360.us> wrote:

> Just had a thought:  Could it be that my Mac Java version is too new?
>
> Mac:
> $ java -version
> java version "1.8.0_60"
> Java(TM) SE Runtime Environment (build 1.8.0_60-b27)
> Java HotSpot(TM) 64-Bit Server VM (build 25.60-b23, mixed mode)
>
> EMR node:
> $ java -version
> java version "1.7.0_71"
> Java(TM) SE Runtime Environment (build 1.7.0_71-b14)
> Java HotSpot(TM) 64-Bit Server VM (build 24.71-b01, mixed mode)
>
> I may try a switching to 1.7 later and report back.
>
> On Sun, Nov 1, 2015 at 9:24 AM, Steve Terrell <sterrell@oculus360.us>
> wrote:
>
>> Thanks, but I'm trying to run remotely.  I'm sure my /etc/hosts is fine
>> as I can ssh and "telnet <host> <ip>" OK.
>>
>> On Sun, Nov 1, 2015 at 9:21 AM, Steve Terrell <sterrell@oculus360.us>
>> wrote:
>>
>>> Thank you, but I'm sure this is not the case as I can easily run
>>> Squirrel client on my mac and query an older version of Phoenix (on another
>>> cluster) via port 2181.
>>>
>>> On Sat, Oct 31, 2015 at 2:21 PM, Naor David <tron47@gmail.com> wrote:
>>>
>>>> run netstat -a 1 | grep "SYN_SENT" and check if port 2181 is blocked in
>>>> your network..
>>>>
>>>> On Sat, Oct 31, 2015 at 8:00 PM, Steve Terrell <sterrell@oculus360.us>
>>>> wrote:
>>>>
>>>>> OK, did some more troubleshooting.  Still can't run sqlline.py from my
>>>>> macbook laptop.  Still hangs.
>>>>>
>>>>> My HBase cluster is an Amazon EMR, and I can run sqlline.py from any
>>>>> of nodes in the cluster, be they master, core, or task nodes.
>>>>>
>>>>> So maybe it's not so much a remote host issue but a problem with my
>>>>> Mac or some kind of Amazon EMR security issue.  But, I've tried opening
up
>>>>> all ports just in case something other than 2181 is required, and no
luck.
>>>>>
>>>>> Has anyone run this version of Phoenix on EMR and been able to use
>>>>> sqlline.py or SQuirreL client remotely from outside of AWS's private
>>>>> network?
>>>>>
>>>>> (My ultimate goal was to get SQuirreL working, but though sqlline.py
>>>>> would be an easier problem to tackle.  SQuirreL is getting timeouts which
I
>>>>> suspect are due to the same hanging that I see with sqlline.py.)
>>>>>
>>>>> Thanks,
>>>>>     Steve
>>>>>
>>>>> On Wed, Oct 28, 2015 at 5:04 PM, Steve Terrell <sterrell@oculus360.us>
>>>>> wrote:
>>>>>
>>>>>> Yes, I can:
>>>>>>
>>>>>> $ telnet *<my master ip>* 2181
>>>>>> Trying 54.174.32.95...
>>>>>> Connected to *<my master host name>*.
>>>>>> Escape character is '^]'.
>>>>>>
>>>>>> Thanks,
>>>>>>     Steve
>>>>>>
>>>>>> On Wed, Oct 28, 2015 at 4:48 PM, Alok Singh <alok@cloudability.com>
>>>>>> wrote:
>>>>>>
>>>>>>> It looks like a zookeeper node is also on the master. Can you
>>>>>>> connect to <master ip> on port 2181 from the machine that
you are running
>>>>>>> sqlline.py on?
>>>>>>>
>>>>>>> Alok
>>>>>>>
>>>>>>> Alok
>>>>>>>
>>>>>>> alok@cloudability.com
>>>>>>>
>>>>>>> On Wed, Oct 28, 2015 at 2:23 PM, Steve Terrell <
>>>>>>> sterrell@oculus360.us> wrote:
>>>>>>>
>>>>>>>> I can get "sqlline.py localhost" to work fine from the master
node.
>>>>>>>>
>>>>>>>> However, when I try to run it remotely, all I get is this:
>>>>>>>>
>>>>>>>> java -cp "*<my path>*/phoenix-4.6.0-HBase-0.98-client.jar"
>>>>>>>> -Dlog4j.configuration=file:*<my path>*/log4j.properties
>>>>>>>> sqlline.SqlLine -d org.apache.phoenix.jdbc.PhoenixDriver
-u jdbc:phoenix:*<my
>>>>>>>> master ip*>:2181:/hbase -n none -p none --color=true
>>>>>>>> --fastConnect=false --verbose=true --isolation=TRANSACTION_READ_COMMITTED
>>>>>>>>
>>>>>>>> Setting property: [isolation, TRANSACTION_READ_COMMITTED]
>>>>>>>> issuing: !connect jdbc:phoenix:*<my master ip*>:2181:/hbase
none
>>>>>>>> none org.apache.phoenix.jdbc.PhoenixDriver
>>>>>>>> Connecting to jdbc:phoenix:*<my master ip*>:2181:/hbase
>>>>>>>> 15/10/28 15:29:44 WARN util.NativeCodeLoader: Unable to load
>>>>>>>> native-hadoop library for your platform... using builtin-java
classes where
>>>>>>>> applicable
>>>>>>>>
>>>>>>>> And then it just hangs.  Has anyone else come across this?
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>>     Steve
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

Mime
View raw message