[
https://issues.apache.org/jira/browse/CAMEL-9060?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ioannis Alexandrakis updated CAMEL-9060:
----------------------------------------
Attachment: jetty-lb-test.zip
First test is OK, when two nodes are up.
Second test fails because the first host to be tried is down, the second one is
up, but the response is not consistent.
> Load balancing http endpoints with camel and jetty9 fails to correctly bridge
> different endpoints (case here: first host is down)
> ---------------------------------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-9060
> URL: https://issues.apache.org/jira/browse/CAMEL-9060
> Project: Camel
> Issue Type: Bug
> Components: camel-jetty
> Affects Versions: 2.15.2, 2.16
> Environment: OS X -yosemite-, java 1.8.0_45-b14
> Reporter: Ioannis Alexandrakis
> Attachments: jetty-lb-test.zip
>
>
> While trying to upgrade an existing http loadbalancer (failover scenario in
> my case) from camel 2.14.x to 2.15.x (I tried in karaf4, so jetty9 was the
> most sensible option instead of 8), I stumbled upon unexpected behavior.
> I realized that, while it was working correctly when both nodes where up and
> serving requests, when the first node was not available (and should
> transparently failover to the second one), I was i) getting the request as
> response, then ii) a consequent request was going to the second server and
> there was a mix of responses. On my development environment I was frequently
> getting "IllegalStateException: AsyncContext completed" between the requests,
> too. The problem was intermittent and I was randomly getting the correct
> response (every 2 or 3 requests)
> If jetty 8 is used, everything is OK. The problem appears with camel-jetty9
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
