Mailing-List: contact user-help@ignite.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@ignite.apache.org
Date: Wed, 4 May 2016 04:32:24 -0700 (PDT)
From: Denis Magda <dmagda@gridgain.com>
To: user@ignite.apache.org
Message-ID: <1462361544380-4757.post@n6.nabble.com>
In-Reply-To: <1462348109910-4752.post@n6.nabble.com>
References: <1462348109910-4752.post@n6.nabble.com>
Subject: Re: Stopping the node in order to prevent cluster wide instability.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
archived-at: Wed, 04 May 2016 11:47:01 -0000

Hi,=20

Please properly subscribe to the user list (this way we will not have to
manually approve your emails). All you need to do is send an email to =C3=
=AC
user-subscribe@ignite.apache.org=C3=AE and follow simple instructions in th=
e
reply.

Upon your questions. Most likely the node, that was stopped, became
segmented: was kicked out of the topology by some reason. Look for "local
node segmented" message is the log.

More often the reason is either is a slow network connection when a node
can't reply on a message during IgniteConfiguration.failureDetectionTimeout
or a long GC pauses.

I would suggest that there are no long GC pauses first. Refer to this page
for more details on how to gather GC logs [1]. If you see pauses bigger tha=
n
10 secs (default value of IgniteConfiguration.failureDetectionTimeout) then
this is the reason why the node was segmented and you have to tune Java hea=
p
[2] and/or your app.

[1]
https://apacheignite.readme.io/v1.5/docs/jvm-and-system-tuning#section-deta=
iled-garbage-collection-stats
[2]
https://apacheignite.readme.io/v1.5/docs/jvm-and-system-tuning#jvm-tuning-f=
or-clusters-with-on_heap-caches

--
Denis


--
View this message in context: http://apache-ignite-users.70518.x6.nabble.co=
m/Stopping-the-node-in-order-to-prevent-cluster-wide-instability-tp4752p475=
7.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.