You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ignite.apache.org by Roman Bielik <ro...@openmindnetworks.com> on 2016/12/21 14:07:42 UTC

Ignite questions

Hi,

I started investigating Apache Ignite as a proof-of-concept for production
cluster-wide key-value store and distributed counters.
I run several tests using REST API on a 3 node cluster; 1 backup; cacheMode
Partitioned; synchronization mode FULL_SYNC.

Questions:
1. When the Ignite instance is idle for couple of minutes I get huge 10
seconds delay for get/put operations at the beginning. Then all the other
requests are responded within 5 mili-seconds at most. All operations are
successful though even those with long response time.

The 10 seconds must be some configuration default, because it is too
precise to be random.
Am I doing something wrong or is this a "normal" warm-up behavior?

2. The REST API seems to lack operation for reseting and/or removing a
counter (atomic long).
Is it a known issue? Is there some workaround to do so?


Thank you,
Regards,
Roman

-- 

<http://www.openmindnetworks.com>
<https://www.linkedin.com/company/openmind-networks>  
<https://twitter.com/Openmind_Ntwks>  <http://www.openmindnetworks.com>

Re: Ignite questions

Posted by Roman <ro...@openmindnetworks.com>.
Hi,

thank you for a quick response.

1. I've found out that the problem was in two network interfaces that I was
using.
All 3 nodes are visible only on one of the interfaces. Interestingly some of
the communication went through (I did see 3 nodes in the topology), some
obviously did not and the delay was caused by:

[09:28:10,129][WARNING][rest-#69%null%][TcpCommunicationSpi] Connect timed
out (consider increasing 'failureDetectionTimeout' configuration property)
[addr=XXXXXX/10.0.2.1:47100, failureDetectionTimeout=10000]

I'm now explicitly setting the localAddress via TcpCommunicationSpi bean and
it seems to have resolved the problem.
Sorry for the false alarm.

2. OK, I will post this as a feature request JIRA ticket.

Thank you,
Best regards,
Roman





--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Ignite-questions-tp9679p9704.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: Ignite questions

Posted by vkulichenko <va...@gmail.com>.
Hi Roman,

1. I don't think this is normal behavior, but I don't think there is any
hardcoded value like this either. Any idea what these 10 seconds are spent
for? Did you do any debugging?

2. I don't there are plans for this. Feel free to create a JIRA ticket (you
can even contribute the implementation yourself).

-Val



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Ignite-questions-tp9679p9684.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.