[
https://issues.apache.org/jira/browse/IGNITE-11871?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Anton Dmitriev updated IGNITE-11871:
------------------------------------
Description:
TensorFlow cluster manager requires NodeId to be resolved into IP address or
hostname to pass the address/name to TensorFlow worker. Currently, it uses
strategy "return first" and returns the first available address/name. As a
result of that, in the case when the server has more than one interface cluster
resolver might work incorrectly and return different addresses/names for the
same server.
To fix this problem we need to update
[TensorFlowServerAddressSpec|https://github.com/apache/ignite/blob/master/modules/tensorflow/src/main/java/org/apache/ignite/tensorflow/cluster/spec/TensorFlowServerAddressSpec.java]
so that it returns the same address/name for the same server all the time. If
a server has multiple network interfaces we need to find a "GCD", a network
with all Ignite nodes.
was:
TensorFlow cluster manager requires NodeId to be resolved into IP address or
hostname to pass the address/name to TensorFlow worker. Currently, it uses
strategy "return first" and returns the first available address/name. As a
result of that, in the case when the server has more than one interface cluster
resolver might work incorrectly and return different addresses/names for the
same server.
To fix this problem we need to update TensorFlowServerAddressSpec so that it
returns the same address/name for the same server all the time. If a server has
multiple network interfaces we need to find a "GCD", a network with all Ignite
nodes.
> [ML] IP resolver in TensorFlow cluster manager doesn't work properly
> --------------------------------------------------------------------
>
> Key: IGNITE-11871
> URL: https://issues.apache.org/jira/browse/IGNITE-11871
> Project: Ignite
> Issue Type: Bug
> Components: ml
> Affects Versions: 2.7
> Reporter: Anton Dmitriev
> Assignee: Anton Dmitriev
> Priority: Major
>
> TensorFlow cluster manager requires NodeId to be resolved into IP address or
> hostname to pass the address/name to TensorFlow worker. Currently, it uses
> strategy "return first" and returns the first available address/name. As a
> result of that, in the case when the server has more than one interface
> cluster resolver might work incorrectly and return different addresses/names
> for the same server.
> To fix this problem we need to update
> [TensorFlowServerAddressSpec|https://github.com/apache/ignite/blob/master/modules/tensorflow/src/main/java/org/apache/ignite/tensorflow/cluster/spec/TensorFlowServerAddressSpec.java]
> so that it returns the same address/name for the same server all the time.
> If a server has multiple network interfaces we need to find a "GCD", a
> network with all Ignite nodes.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
