Add node automatically to Ignite

[Amit Pandey]

Hi,

I have a cluster of Ignite nodes running with persistence with docker. So we activate the cluster programmatically the first time and it works fine.

However , when a node fails and we have to add a new node in its place we are seeing that its not joining the cluster.

is there anyway to fix this ?

regards

[Amit Pandey]

Just to add that the issue is that all our caches are replicated and it seems that when we hit the new node we dont get data replicated at all.

Regards

On Wed, Nov 25, 2020 at 1:29 AM Amit Pandey <[hidden email]> wrote:

Hi,

I have a cluster of Ignite nodes running with persistence with docker. So we activate the cluster programmatically the first time and it works fine.

However , when a node fails and we have to add a new node in its place we are seeing that its not joining the cluster.

is there anyway to fix this ?

regards

[stephendarlington]

Once you’ve activated your cluster, you need to add and remove nodes from the baseline manually. The most common method is "control.sh —baseline add/remove" but there’s also an API.

You could also consider baseline topology auto-adjustment, but you need to be very careful that you don’t move data around unnecessarily.

Both are documented here:

https://ignite.apache.org/docs/latest/clustering/baseline-topology#baseline-topology-in-persistent-clusters

Regards,

Stephen

On 24 Nov 2020, at 20:01, Amit Pandey <[hidden email]> wrote:

Just to add that the issue is that all our caches are replicated and it seems that when we hit the new node we dont get data replicated at all.

Regards

On Wed, Nov 25, 2020 at 1:29 AM Amit Pandey <[hidden email]> wrote:

Hi,

I have a cluster of Ignite nodes running with persistence with docker. So we activate the cluster programmatically the first time and it works fine.

However , when a node fails and we have to add a new node in its place we are seeing that its not joining the cluster.

is there anyway to fix this ?

regards

[Amit Pandey]

I want to do a check then so,

if I do , the below when cluster is activated then if I add or remove nodes after a given time Ignite will take care of the balancing ? Is that understanding correct? In our case its not possible to use the control script

ignite.cluster().baselineAutoAdjustEnabled(true);

ignite.cluster().baselineAutoAdjustTimeout(30000);

On Wed, Nov 25, 2020 at 3:04 PM Stephen Darlington <[hidden email]> wrote:

Once you’ve activated your cluster, you need to add and remove nodes from the baseline manually. The most common method is "control.sh —baseline add/remove" but there’s also an API.

You could also consider baseline topology auto-adjustment, but you need to be very careful that you don’t move data around unnecessarily.

Both are documented here:

https://ignite.apache.org/docs/latest/clustering/baseline-topology#baseline-topology-in-persistent-clusters

Regards,

Stephen

On 24 Nov 2020, at 20:01, Amit Pandey <[hidden email]> wrote:

Just to add that the issue is that all our caches are replicated and it seems that when we hit the new node we dont get data replicated at all.

Regards

On Wed, Nov 25, 2020 at 1:29 AM Amit Pandey <[hidden email]> wrote:

Hi,

I have a cluster of Ignite nodes running with persistence with docker. So we activate the cluster programmatically the first time and it works fine.

However , when a node fails and we have to add a new node in its place we are seeing that its not joining the cluster.

is there anyway to fix this ?

regards

[stephendarlington]

Yes.

On 25 Nov 2020, at 12:31, Amit Pandey <[hidden email]> wrote:

I want to do a check then so,

if I do , the below when cluster is activated then if I add or remove nodes after a given time Ignite will take care of the balancing ? Is that understanding correct? In our case its not possible to use the control script

ignite.cluster().baselineAutoAdjustEnabled(true);

ignite.cluster().baselineAutoAdjustTimeout(30000);

On Wed, Nov 25, 2020 at 3:04 PM Stephen Darlington <[hidden email]> wrote:

Once you’ve activated your cluster, you need to add and remove nodes from the baseline manually. The most common method is "control.sh —baseline add/remove" but there’s also an API.

You could also consider baseline topology auto-adjustment, but you need to be very careful that you don’t move data around unnecessarily.

Both are documented here:

https://ignite.apache.org/docs/latest/clustering/baseline-topology#baseline-topology-in-persistent-clusters

Regards,

Stephen

On 24 Nov 2020, at 20:01, Amit Pandey <[hidden email]> wrote:

Just to add that the issue is that all our caches are replicated and it seems that when we hit the new node we dont get data replicated at all.

Regards

On Wed, Nov 25, 2020 at 1:29 AM Amit Pandey <[hidden email]> wrote:

Hi,

I have a cluster of Ignite nodes running with persistence with docker. So we activate the cluster programmatically the first time and it works fine.

However , when a node fails and we have to add a new node in its place we are seeing that its not joining the cluster.

is there anyway to fix this ?

regards