Re: getOrCreateCache hang

[Matt Warner <ma...@warnertechnology.com>]

Nikolai, I feel disappointed with how this email chain seems to be going. I
spent time and effort to assemble sample code that exhibits the problem
consistently only to hear that it doesn't seem like you're even using that
code. That leaves me wondering what you were actually testing, when you
told me you couldn't reproduce this. It does not sound like you were using
the code I supplied.

I also sent thread dumps and logs early on in this process, and the current
thread dumps show the same as before—getOrCreateCache.

This problem does seem tied to the CacheJdbcStoreSessionListener data
source, but I've so far been unable to pin it down beyond that. No doubt
it's something simple I'm doing wrong, but whatever it is still eludes me.

Matt

On Tue, Feb 28, 2017 at 2:14 AM, Nikolai Tikhonov <nt...@apache.org>
wrote:

> I am not able reproduce exactly your test (do not have Postgre instance),
>  but I do not get hangs on start caches. I suppose that issue related with
> your setup and for getting cause of problem I need full logs and thread
> dumps (by kill -3 PID or jstack) from all nodes. Could you please provide
> this data?
>
> On Mon, Feb 27, 2017 at 10:36 PM, Matt Warner <ma...@warnertechnology.com>
> wrote:
>
>> Something's not right. I'm seeing this as 100% reproducible on two
>> different OSes.
>>
>> Are you running "mvn package" on both testIgnite1 and testIgnite2 and
>> then starting the shaded JARs simultaneously from two different shell
>> windows using "java-jar ignite?...-standalone.jar"?
>>
>> Matt
>>
>> On Feb 27, 2017, at 8:58 AM, Nikolai Tikhonov <nt...@apache.org>
>> wrote:
>>
>> I was not able reproduce it. Could you share full logs and thread dumps
>> from all nodes?
>>
>> On Mon, Feb 27, 2017 at 7:45 PM, Matt Warner <ma...@warnertechnology.com>
>> wrote:
>>
>>> Using the test code I sent previously, I changed the ports to be a range
>>> but both clients still deadlock on getOrCreateCache.
>>>
>>> Are you able to reproduce this using the test code I sent?
>>>
>>> Matt
>>>
>>>
>>> On Mon, Feb 27, 2017 at 3:02 AM, Nikolai Tikhonov <nt...@apache.org>
>>> wrote:
>>>
>>>> Hi Matt!
>>>>
>>>> Try to change ipFinder.setAddresses(Arrays.asList("127.0.0.1:47500")); to
>>>> ipFinder.setAddresses(Arrays.asList("127.0.0.1:47500..47505"));
>>>>
>>>> On Fri, Feb 24, 2017 at 4:00 PM, Matt Warner <matt@warnertechnology.com
>>>> > wrote:
>>>>
>>>>> Hi Nikolai.
>>>>>
>>>>> I discovered the reason the two applications weren't seeing each other
>>>>> was resolved by adding an explicit port number (Arrays.asList("
>>>>> 127.0.0.1:47500")). However, the two still deadlock when running
>>>>> concurrently.
>>>>>
>>>>> The latest test shows one application blocked in getOrCreateCache, the
>>>>> other blocked in Ignition.start. As soon as I kill the process stuck in
>>>>> Iginition.start the other process continues successfully. I've attached the
>>>>> latest test code.
>>>>>
>>>>> Any ideas?
>>>>>
>>>>> Matt
>>>>>
>>>>> On Wed, Feb 22, 2017 at 10:30 AM, Matt Warner [via Apache Ignite
>>>>> Users] <[hidden email]
>>>>> <http:///user/SendEmail.jtp?type=node&node=10866&i=0>> wrote:
>>>>>
>>>>>> One other observation: with the third application acting as just the
>>>>>> server, and just one of the clients running, there is no issue. Only when
>>>>>> there are multiple clients do I get the deadlock on getOrCreateCache.
>>>>>>
>>>>>> Matt
>>>>>>
>>>>>> ------------------------------
>>>>>> If you reply to this email, your message will be added to the
>>>>>> discussion below:
>>>>>> http://apache-ignite-users.70518.x6.nabble.com/getOrCreateCa
>>>>>> che-hang-tp10737p10811.html
>>>>>> To unsubscribe from getOrCreateCache hang, click here.
>>>>>> NAML
>>>>>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>>>>>
>>>>>
>>>>>
>>>>> *testIgnite.tar.gz* (14K) Download Attachment
>>>>> <http://apache-ignite-users.70518.x6.nabble.com/attachment/10866/0/testIgnite.tar.gz>
>>>>>
>>>>> ------------------------------
>>>>> View this message in context: Re: getOrCreateCache hang
>>>>> <http://apache-ignite-users.70518.x6.nabble.com/getOrCreateCache-hang-tp10737p10866.html>
>>>>>
>>>>> Sent from the Apache Ignite Users mailing list archive
>>>>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>>>>
>>>>
>>>>
>>>
>>
>

Re: getOrCreateCache hang

[bintisepaha <bi...@tudor.com>]

Matt, I have not tried your code (not a committer to the community), but what
happens when you use Ignite.cache(), if the cache was already created on
server startup.



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/getOrCreateCache-hang-tp10737p11065.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: getOrCreateCache hang

[vkulichenko <va...@gmail.com>]

Hi Matt,

First of all, this is a community forum and it's not a place for
disappointment :) Nikolai is not obligated to help and I'm sure he is trying
his best when doing this.

I managed to reproduce the hang though. It happens simply because you have
two projects (why?) that don't share classes. In particular, when
testIgnite1 starts the cache, it provides its own store factory that doesn't
exist in testIgnite2, thus the failure. Actually, if you switch off quiet
mode (set -DIGNITE_QUIET=false), you will see ClassNotFoundException right
away. Such classes should be shared across all nodes in topology. If I start
two nodes from the same project, it works fine.

-Val



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/getOrCreateCache-hang-tp10737p11097.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.