Re: Using Persistent ignite queues

[Denis Mekhanikov]

Hi!

IgniteQueue is stored in the atomics cache, which is called 
ignite-sys-atomic-cache@default-ds-group by default.
This cache is stored in the default data region, so in order to make it 
persisted, you need to make the default data region persisted using 
DataStorageConfiguration#defaultDataRegionConfiguration property.

The issue with the deadlock seems similar to the following one: 
https://issues.apache.org/jira/browse/IGNITE-10250

You can try a nightly build, where this issue is fixed: 
https://ci.ignite.apache.org/viewLog.html?buildId=lastSuccessful=Releases_NightlyRelease_RunApacheIgniteNightlyRelease=artifacts=1

Denis
On 10 Aug 2018, 13:32 +0300, dkol , wrote:
> Hi arunkjn
>
> were you able to resolve this issue ?
>
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Using Persistent ignite queues

[dkol]

Hi arunkjn

were you able to resolve this issue ?




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Using Persistent ignite queues

[arunkjn]

Hi Alex,

I cannot get it to reproduce in my local environment yet. I will get back to
you with thread dumps as soon as I encounter this issue again.



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Using Persistent ignite queues

[aealexsandrov]

Hi,

Could you please provide next for investigation:

1)Thread dump that you should create at the moment when cluster hung.
2)Code of your service
3)Logs of the cluster nodes.
4)The configuration of the clusters.

BR,
Andrei



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Using Persistent ignite queues

[arunkjn]

Hi,

I have a scenario where I use a persistent ignite queue across cluster
nodes. 

This is my setup using ignite 2.5-

Few server nodes which hosts some caches in persistance mode and a cluster
singleton ignite service. This ignite service upon initialization,
initializes two persistent queues.

Rest of the nodes in cluster are client nodes. They either put data in queue
or consume data from queue (depending on their role).  Basically I am using
Ignite queue as a worker queue, one party fills the queue and another party
empties it.

I have observed that sometimes the queue gets stuck, as in the queue.take()
method on one node keeps blocking even if there are items in the queue. This
issue does not go away until I restart the whole cluster from fresh state.

I looked into the /etc/ignite/peristance/1 folder (1 is the consistent node
id for one of the nodes). It only contains folders to store the caches. I
cannot see any reference of queue getting saved. I am missing something
here?

Thanks,
Arun



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/