[
https://issues.apache.org/jira/browse/SYNCOPE-456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13837254#comment-13837254
]
Hadrian Zbarcea commented on SYNCOPE-456:
-----------------------------------------
Mystery solved. I suspect other devs in the US might have experienced the same
thing. My isp is verizon (fios) and I get a dynamic ip and inherit the dns
servers configured by the isp. I do not define my own. Verizon has a 'feature'
in the dns server where, if you query for a non-fqdn host it actually resolves
it to one of its own dns search servers. As such, on my system, I was never
getting the {{java.net.UnknownHostException: not-existing}} because the dns was
resolving the "non-existing" host to 199.101.28.20 (which is actually
http://search.dnsassist.verizon.net/).
{code}
hadrian@rem:syncope$ nslookup non-existing
Server: 127.0.1.1
Address: 127.0.1.1#53
Non-authoritative answer:
Name: non-existing
Address: 199.101.28.20
{code}
Next syncope attempted to connect to the (assumed to be an smtp) server. Due to
the default configuration (see SYNCOPE-457) the result was an infinite wait and
hence a hang in that thread.
Using an FQDN for the 'non-existing' host makes the dns behave as expected and
solves the test problem. See patch. Patch contains some extra entries in
.gitignore to ignore eclipse generated artifacts.
> TaskTestITCase failures
> -----------------------
>
> Key: SYNCOPE-456
> URL: https://issues.apache.org/jira/browse/SYNCOPE-456
> Project: Syncope
> Issue Type: Bug
> Components: core
> Affects Versions: 1.2.0
> Reporter: Hadrian Zbarcea
> Assignee: Hadrian Zbarcea
> Attachments: syncope-456.patch
>
>
> Both issueSYNCOPE81 and issueSYNCOPE86 fail on my system. The symptom is that
> TaskController.execute hangs (the line below is the last in the log) and then
> there is a timeout on the client side.
> 22:46:09.087 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler
> - Before TaskController.execute([101, false])
> I am not sure if anybody saw this, if so please add a comment. I reproduce
> the issue consistently on my Ubuntu laptop with both sun and openjdk 6.
> Others don't see it in an environment similar to mine. I also saw it on a
> CentOS server. I cannot reproduce it on osx though. It is quite likely
> related to the environment, but nevertheless syncope should be more robust
> and handle errors gracefully in a misconfigured environment.
> Since I have environments that both work and don't, I will try to dig deeper,
> compare the two and see what the culprit is.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
