[
https://issues.apache.org/jira/browse/CALCITE-924?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14962579#comment-14962579
]
Josh Elser commented on CALCITE-924:
------------------------------------
bq. Can you define "unique"?
Good point. I think we'd want "unique" to be "reasonably likely that a single
client (JVM) would only see unique connection IDs".
PID of the JVM, millis since epoch and IP would likely be sufficient to
uniquely identify one "client". Then, some random number or some counter for
uniqueness among multiple connections from that "client". Very similar to what
Nick proposed initially (probably because my inspiration would have also come
from HBase, but whatever).
> Ensure connectionIDs are unique
> -------------------------------
>
> Key: CALCITE-924
> URL: https://issues.apache.org/jira/browse/CALCITE-924
> Project: Calcite
> Issue Type: Improvement
> Components: avatica
> Reporter: Josh Elser
> Assignee: Julian Hyde
>
> Follow-up work from CALCITE-906:
> Some recommendations from [~ndimiduk] from HBase-landia:
> {quote}
> We'll need unique connection ID's as well. HBase uses a combination of a
> unique hash per client + random number. Maybe we can do similarly for
> connections, and with a connection hash plus random number for statements.
> https://github.com/apache/hbase/blob/master/hbase-client/src/main/java/org/apache/hadoop/hbase/client/PerClientRandomNonceGenerator.java
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
