Amazon AWS EC2 settings

[ghughal <ba...@gmail.com>]

We are running ignite on AWS EC2 instance. We are not using S3 based
discovery but using zookeeper based discovery and it seems to work fine.

While reading some of ignite documentation I came across this that suggests
using "significantly greater than the default" value for socket timeout for
EC2 instances. 

https://ignite.apache.org/releases/mobile/org/apache/ignite/spi/discovery/tcp/TcpDiscoverySpi.html#setSocketTimeout(long)

As per same documentation, if I set socket timeout then
failureDetectionTimeout is ignored and I need to set ackTimeout,
maxAckTimeout, heartbeatFrequency, reconnectCount etc. explicitly. 

Do you have any recommendation on what values we should be using for these
settings in AWS? Or do you have any recommendation value for
failureDetectionTimeout for EC2 instances?

With default settings, once in a while under heavy load, we get network
split with errors similar to this:





--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Amazon-AWS-EC2-settings-tp10060.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: Amazon AWS EC2 settings

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

First of all, I would check that there are no memory issue under load. This
(as message states) is likely to happen when one or more nodes suffer from
long GC pauses. If this is not the case, I would set failure detection
timeout to several seconds and then try several values to check which one
works.

-Val



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Amazon-AWS-EC2-settings-tp10060p10062.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.