[
https://issues.apache.org/jira/browse/NIFI-2459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15434546#comment-15434546
]
ASF GitHub Bot commented on NIFI-2459:
--------------------------------------
GitHub user ijokarumawak opened a pull request:
https://github.com/apache/nifi/pull/927
NIFI-2459: HTTP Site-to-Site can't update remote peer topology if the
"bootstrap url" node is down
This PR migrates existing code which handles the situation, from
EndpointConnectionPool to PeerSelector, so that both RAW and HTTP transport
protocol has the same capability.
Added a unit test to confirm this scenario.
Also did an IT test using running NiFi cluster:
Used a 3 node cluster. Setup S2S specifying node 2 as the bootstrap node,
then stop node 2. Confirm data is sent to only node 1 and 3. Remote peer status
can still be refreshed. Then add Node 2 back, Confirm data is sent to node 1, 2
and 3.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/ijokarumawak/nifi nifi-2459
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/nifi/pull/927.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #927
----
commit ca3a365d4ec649f7d4b5742c37fb6c250bb73540
Author: Koji Kawamura <[email protected]>
Date: 2016-08-24T08:59:25Z
NIFI-2459: Site-to-Site bootstrap node failure
Refresh remote peer statuses even if the bootstrap node goes down.
Migrate existing code which handles the situation from
EndpointConnectionPool to PeerSelector, so that both RAW and HTTP
transport protocol has the same capability.
----
> HTTP Site-to-Site can't update remote peer topology if the "bootstrap url"
> node is down
> ---------------------------------------------------------------------------------------
>
> Key: NIFI-2459
> URL: https://issues.apache.org/jira/browse/NIFI-2459
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Affects Versions: 1.0.0
> Reporter: Koji Kawamura
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> To establish a Site-to-Site protocol, a client NiFi uses the remote NiFi API
> url that is specified as 'URL' at the creation of a Remote Process Group
> (let's call it as 'bootstrap url').
> Currently, HTTP Site-to-Site keeps using the same 'bootstrap url', even if
> the remote node goes down. If the bootstrap node went down, Site-to-Site data
> transfer will keep working, but it can't update remote peer topology.
> RAW socket version cares about this scenario by asking updated peer
> information with every nodes that it has connected before, until it gets a
> successful response.
> HTTP version should follow how RAW version handles this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
