[
https://issues.apache.org/jira/browse/CASSANDRA-12039?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15391866#comment-15391866
]
Sam Tunnicliffe commented on CASSANDRA-12039:
---------------------------------------------
Thanks, the updated patch is looking good. If you amend slightly the new log
statement in {{SIM}} to include the value being passed in, we can properly
check that in a dtest. Otherwise, modulo the couple of nits below I think we'll
be good to go once we get some dtests in place.
One thing that we should note is that when pre-join tasks are executed as a
result of manually joining (i.e. start in write survey mode, then join via
nodetool/JMX), the value of the bootstrap flag is always false. Obviously, this
might not always be accurate as a node can perfectly well start up in survey
mode but also perform bootstrap before an operator joins it. Short of exposing
the overload of {{joinRing}} and extending the nodetool join command, I'm not
sure there's any practical solution to this. However, I guess this is quite a
niche scenario, so I suggest that clearly documenting this limitation in
{{NEWS.txt}} is sufficient. Would you mind adding an entry to {{NEWS.txt}}
please?
Nits:
The filter condition in {{StorageService::executePreJoinTasks}} can be
streamlined to:
{code}
.filter(cfs ->
Schema.instance.getUserKeyspaces().contains(cfs.keyspace.getName()))
{code}
{{SIM::executeAllBlocking}} handles null {{Callable}} correctly, so the default
impl of {{getPreJoinTask}} can just return {{null}} rather than a task which
returns {{null}}.
There's a couple of other minor formatting things, but I can fix those on
commit.
> Add an index callback to be notified post bootstrap and before joining the
> ring
> -------------------------------------------------------------------------------
>
> Key: CASSANDRA-12039
> URL: https://issues.apache.org/jira/browse/CASSANDRA-12039
> Project: Cassandra
> Issue Type: New Feature
> Reporter: Sergio Bossa
> Assignee: Sergio Bossa
>
> Custom index implementations might need to be notified when the node finishes
> bootstrapping in order to execute some blocking tasks before the node itself
> goes into NORMAL state.
> This is a proposal to add such functionality, which should roughly require
> the following:
> 1) Add a {{getPostBootstrapTask}} callback to the {{Index}} interface.
> 2) Add an {{executePostBootstrapBlockingTasks}} method to
> {{SecondaryIndexManager}} calling into the previously mentioned callback.
> 3) Hook that into {{StorageService#joinTokenRing}}.
> Thoughts?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
