Hi,
Currently node id and name change at each restart. Is it possible to store
them (at least id) and reuse that node id? We are planning to monitor ES
cluster, nodes, indices and etc. However, since node id change every time it
is restarted, it is not possible to track a node.
Node name can be set by setting name setting, id is always auto generated (its highly important to make sure its unique). It can be allowed to be set, but I wonder if a different way to try and attack this is not better, for example, I have been thinking that node level APIs should allow to specify the name (or even IP address) for example.
On Friday, June 10, 2011 at 3:20 PM, Mustafa Sener wrote:
Hi,
Currently node id and name change at each restart. Is it possible to store them (at least id) and reuse that node id? We are planning to monitor ES cluster, nodes, indices and etc. However, since node id change every time it is restarted, it is not possible to track a node.
Hi,
I understand your concerns about auto generated id. However, once it is auto
generated why we don't save it to meta data and reuse that auto generated id
instead of generating new one. In this way we can be sure that it is unique
since it is generated randomly once and we can refer to that node with that
id for subsequent restarts.
Node name can be set by setting name setting, id is always auto
generated (its highly important to make sure its unique). It can be allowed
to be set, but I wonder if a different way to try and attack this is not
better, for example, I have been thinking that node level APIs should allow
to specify the name (or even IP address) for example.
On Friday, June 10, 2011 at 3:20 PM, Mustafa Sener wrote:
Hi,
Currently node id and name change at each restart. Is it possible to store
them (at least id) and reuse that node id? We are planning to monitor ES
cluster, nodes, indices and etc. However, since node id change every time it
is restarted, it is not possible to track a node.
Since then we need to somehow identify a "machine" or an instance (basically a JVM) and then it becomes much more complex. I was thinking about stats reports, you can use things like IP, or hostname, to report it, or, node name if needed, which you can set.
On Monday, June 13, 2011 at 9:07 AM, Mustafa Sener wrote:
Hi,
I understand your concerns about auto generated id. However, once it is auto generated why we don't save it to meta data and reuse that auto generated id instead of generating new one. In this way we can be sure that it is unique since it is generated randomly once and we can refer to that node with that id for subsequent restarts.
Node name can be set by setting name setting, id is always auto generated (its highly important to make sure its unique). It can be allowed to be set, but I wonder if a different way to try and attack this is not better, for example, I have been thinking that node level APIs should allow to specify the name (or even IP address) for example.
On Friday, June 10, 2011 at 3:20 PM, Mustafa Sener wrote:
Hi,
Currently node id and name change at each restart. Is it possible to store them (at least id) and reuse that node id? We are planning to monitor ES cluster, nodes, indices and etc. However, since node id change every time it is restarted, it is not possible to track a node.
Node name can be set by setting name setting, id is always auto generated (its highly important to make sure its unique). It can be allowed to be set, but I wonder if a different way to try and attack this is not better, for example, I have been thinking that node level APIs should allow to specify the name (or even IP address) for example.
Node name can be set by setting name setting, id is always auto generated (its highly important to make sure its unique). It can be allowed to be set, but I wonder if a different way to try and attack this is not better, for example, I have been thinking that node level APIs should allow to specify the name (or even IP address) for example.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.