Hi,
Regarding the ports, I'm copy-pasting the response provided by Nikolai on
dev list:
Ignite by default using the following ports:11211 - REST, 47100..47200 -
TCP communication, 47500...47600 - Discovery, 49112 - JMX. You can change
the values (see TcpCommunicationSpi#setLocalPort and
Can you pleas tell me what ports should be opened on driver or yarn cluster
machine? this is our production env, I suspect that something need be
configured on firewall ports.
1)
I verified all these steps on my local cluster(3 spark machine[1 master and
2 workers], one yarn cluster machine),
Hello, Nikolai,
YES, I can ping yarn cluster from driver machine.
I try to telnet localhost 50075 on driver machine, but failed, is this root
cause for this issues? thanks!!!
--
View this message in context:
Hi,
Could you make sure that you have access from "driver" machine to machines
on which running YARN? Could you ping them?
On Fri, Aug 12, 2016 at 11:26 AM, percent620 wrote:
> scala> [14:52:04] New version is available at ignite.apache.org: 1.7.0
> 16/08/12 14:52:46 ERROR
scala> [14:52:04] New version is available at ignite.apache.org: 1.7.0
16/08/12 14:52:46 ERROR TcpDiscoverySpi: Failed to reconnect to cluster
(consider increasing 'networkTimeout' configuration property)
[networkTimeout=5000]
16/08/12 15:37:54 ERROR GridClockSyncProcessor: Failed to send time
Get the following error message after a few minutes
scala> val ic = new IgniteContext[Integer,
Integer](sc,"/u01/yueyi/apache-ignite-fabric-1.6.0-bin/config/default-config.xml")
ic: org.apache.ignite.spark.IgniteContext[Integer,Integer] =
org.apache.ignite.spark.IgniteContext@26917c50
scala>
Hello, Nikolai,
I'm a new ignite, can you please provide more detail steps for this? thanks
again!!!
--
View this message in context:
http://apache-ignite-users.70518.x6.nabble.com/Ignite-for-Spark-on-YARN-Deployment-tp6910p6986.html
Sent from the Apache Ignite Users mailing list archive at
>
> 3) I'm running spark-shell on driver machine NOT yarn cluster?
>
ignite-spark starts Ignite client node which should have direct access to
YARN ignite cluster (have network access, open ports and etc).
1) containers logs as below
1、[20:07:14] Ignite node started OK (id=a3a87e37)
[20:07:14] Topology snapshot [ver=1, servers=1, clients=0, CPUs=24,
heap=2.0GB]
[20:07:14] Topology snapshot [ver=2, servers=2, clients=0, CPUs=48,
heap=4.0GB]
[20:07:15] Topology snapshot [ver=3, servers=3, clients=0,
Could you show logs from containers? spark-shell just hangs or print some
logs? Also are you sure that you have direct access to machine on which
running yarn cluster?
On Thu, Aug 11, 2016 at 3:43 PM, percent620 wrote:
> Hello, Nikolai,
> 1、
> Just updated configuration
Hello, Nikolai,
1、
Just updated configuration default-config.xm on
/u01/XXX/apache-ignite-fabric-1.6.0-bin/config/default-config.xm
2)./hdfs dfs -put
/u01/XXX/apache-ignite-fabric-1.6.0-bin/config/default-config.xm
/ignite/release16/apache-ignite-fabric-1.6.0-bin/config/
3)
scala> import
Hi,
You can use hdfs path as
hdfs://your_host:9000/ignite/release16/apache-ignite-fabric-1.6.0-bin/config/default-config.xml.
Or you can just copy this config file on local disk. ;) You should use the
same configuration, but it can be different files with the same content.
On Thu, Aug 11, 2016
Thanks Nikolai very much.
As your request, and changed configuration
/ignite/release16/apache-ignite-fabric-1.6.0-bin/config/default-config.xml
but this file is hfs file
1)
$./hdfs dfs -text
/ignite/release16/apache-ignite-fabric-1.6.0-bin/config/default-config.xml
SLF4J: Class path contains
Great! Ignite YARN cluster successfully started, but ignite spark shell
doesn't see server nodes. You need to configure IP finder, by default YARN
cluster using VmIpFinder, but ignite-spark MulticastIpFinder. Could you
change configuration for YARN cluster (IGNITE_XML_CONFIG=/ignite/
1、Adjusted cluster.properites as below
$cat cluster16.properties
# The number of nodes in the cluster.
IGNITE_NODE_COUNT=4
# The number of CPU Cores for each Apache Ignite node.
IGNITE_RUN_CPU_PER_NODE=1
# The number of Megabytes of RAM for each Apache Ignite node.
IGNITE_MEMORY_PER_NODE=2048
#
Thanks very much for your quick response.!!!
I will attache contain logs for this issues.
Can you please also help me on this issues as below?
http://apache-ignite-users.70518.x6.nabble.com/Embedded-mode-ignite-on-spark-tt6942.html
At 2016-08-11 17:43:40, "Nikolai Tikhonov-2 [via Apache
16 matches
Mail list logo