[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17298146#comment-17298146
]
Andrzej Bialecki commented on SOLR-14749:
-
It looks like the underlying cause of the failures was the same as in
SOLR-15122. I added a Phaser-based mechanism for tests to monitor the changes
in configuration in the {{ContainerPluginsRegistry}}, similar to the one used
in SOLR-15122.
I'm leaving this issue open to see if the fix works on jenkins (local beasting
can't reproduce this failure).
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17298138#comment-17298138
]
ASF subversion and git services commented on SOLR-14749:
Commit 7ada4032180b516548fc0263f42da6a7a917f92b in lucene-solr's branch
refs/heads/master from Andrzej Bialecki
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=7ada403 ]
SOLR-14749: Make sure the plugin config is reloaded on Overseer.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17294211#comment-17294211
]
Tomas Eduardo Fernandez Lobbe commented on SOLR-14749:
--
[~dsmiley] brought up
[here|https://issues.apache.org/jira/browse/SOLR-14663?focusedCommentId=17293955&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17293955]
that {{TestContainerPlugin.testApi}} has been failing intermittently. I
believe the timeline and context of the failures aligns with the commit done in
Oct 21st for this Jira issue
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17229059#comment-17229059
]
ASF subversion and git services commented on SOLR-14749:
Commit bac43093265c56996d35f2d5d9f93c4323a7b7e5 in lucene-solr's branch
refs/heads/master from Adrien Grand
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=bac4309 ]
SOLR-14749: Use h2 instead of h3 so that the javadoc tool doesn't complain
about out-or-sequence headers.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17226705#comment-17226705
]
ASF subversion and git services commented on SOLR-14749:
Commit 0bfa2a690857099aa3a8a5bd90e4d8db89d5ccc0 in lucene-solr's branch
refs/heads/master from Andrzej Bialecki
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=0bfa2a6 ]
SOLR-14749: Restructure the docs + add some examples.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17226674#comment-17226674
]
Andrzej Bialecki commented on SOLR-14749:
-
Thanks [~noble.paul] and [~ilan] for comments and reviews!
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 22h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17226657#comment-17226657
]
ASF subversion and git services commented on SOLR-14749:
Commit bdc6e8247fdb162902c794b73fdc228d526a3a6e in lucene-solr's branch
refs/heads/master from Andrzej Bialecki
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=bdc6e82 ]
SOLR-14749: Provide a clean API for cluster-level event processing.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 21h 40m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=1702#comment-1702
]
Andrzej Bialecki commented on SOLR-14749:
-
This is not strictly speaking a user-level functionality. As we discussed this
offline, it's true that we need to provide a comprehensive documentation for
the whole plugin subsystem, but in my view this falls somewhere between
admin-level and developer-level.
Admin-level docs would include the list of plugin categories, bundled
implementations and how to configure them. {{ClusterSingleton}} is a
developer-level concept, which is not exposed to admins and not configurable
(at least for now). The {{ClusterEventProducer}} implementation and
{{ClusterEventListener}}-s are configurable, and I can add some docs
(somewhere???) but their functionality can only be changed by providing a
different plugin implementation ... which is again a developer-level doc.
I don't think we have any good place to put developer-level documentation yet -
the Ref Guide doesn't seem ideal, its focus is on user- and admin-level
documentation.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 19.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17221797#comment-17221797
]
Noble Paul commented on SOLR-14749:
---
I mean we have 2 new capabilities
* Registering a cluster singleton
* Registering an event listener
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 19.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17221730#comment-17221730
]
Andrzej Bialecki commented on SOLR-14749:
-
[~noble.paul] "this" meaning what part specifically? so far this is an internal
API, unless you mean the ability to define non-API plugins?
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 19.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17221293#comment-17221293
]
Noble Paul commented on SOLR-14749:
---
[~ab], this is a user-facing change, Can you please add add relevant examples
in the description?
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 19h 20m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17218444#comment-17218444
]
ASF subversion and git services commented on SOLR-14749:
Commit 67ecd8ff9ad3640016424ded86bfaaadd815b96d in lucene-solr's branch
refs/heads/master from Andrzej Bialecki
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=67ecd8f ]
SOLR-14749: Improve support for arbitrary container-level plugins. Add
ClusterSingleton
support for plugins that require only one active instance in the cluster.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 18h 10m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17210168#comment-17210168
]
Noble Paul commented on SOLR-14749:
---
Thanks [~ab] . I shall review it soon
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 15.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17210165#comment-17210165
]
Andrzej Bialecki commented on SOLR-14749:
-
[~noble.paul] PR-1964 contains just the {{ClusterSingleton}} part with the
testcase and the changes needed to support loading this type of plugins via
{{CustomContainerPlugins}}.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 15.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17210134#comment-17210134
]
Noble Paul commented on SOLR-14749:
---
[~ab]
The APIs themselves are pretty clean .
Can we have {{ClusterSingleton}} impl with testcase and merge it in a single PR?
That way we can get one thing out of the way soon.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 15h 20m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17210131#comment-17210131
]
Andrzej Bialecki commented on SOLR-14749:
-
[~noble.paul] please see PR-1962 that contains just the proposed APIs, no other
changes (and no implementation).
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 15h 20m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17206584#comment-17206584
]
Noble Paul commented on SOLR-14749:
---
{quote}I had the impression we are having a proper discussion, both here and on
the PR. Your -1 means you have serious technical objections, correct? {quote}
My objections are not on the ticket. I'm objecting to the PR. I see too many
different changes being made and this is not healthy
I request you to make focussed PRs (if required make sub JIRAs) so that others
can make meaningful suggestions
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 14h 50m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17206361#comment-17206361
]
Andrzej Bialecki commented on SOLR-14749:
-
{quote}The PR is too complex.
{quote}
What part is too complex? The proposed interfaces or the proof-of-concept
implementation? The APIs are quite simple IMHO.
{quote}please do not make backward incompatible changes
{quote}
Sure, I reverted that change, although it would have been simple to accommodate
the name change so that we preserve back-compat, and the current name is
illogical - {{/cluster/plugin}} vs. {{/cluster/plugins}} when the location
keeps information about multiple plugins. In any case, it's a trivial change so
let's not focus on this.
{quote}Here is my official -1 on this PR. We would like to have a proper
discussion
{quote}
I had the impression we are having a proper discussion, both here and on the
PR. Your -1 means you have serious technical objections, correct? If so then
please explain in detail your objections to the proposed implementation, having
in mind the goal and scope of this issue. If you have in mind an alternative
approach that can satisfy these goals then I'm all ears.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 14h 20m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17205388#comment-17205388
]
Ishan Chattopadhyaya commented on SOLR-14749:
-
bq. I would like to see a separate PR for ClusterSingleton and commit that
separately.
[~noble.paul], my objections to putting ClusterSingleton in solr-core stands.
But, I shall not veto on these grounds if all of you are convinced that is the
only way forward. Here's another reason why I am not pleased to have it:
https://github.com/apache/lucene-solr/pull/1758#issuecomment-701966468.
Unfortunately, I am very busy at the moment (personal travel and other work)
and won't be able to engage/help in this issue; apologies (will probably have a
more relaxed schedule after 15-20 days).
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 14h 10m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17205364#comment-17205364
]
Noble Paul commented on SOLR-14749:
---
Here is my official -1 on this PR
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 14h 10m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17204769#comment-17204769
]
Noble Paul commented on SOLR-14749:
---
[~ab] please do not make backward incompatible changes without proper
discussions
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 13h 40m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17201299#comment-17201299
]
Noble Paul commented on SOLR-14749:
---
The PR is too complex.
I would like to see a separate PR for {{ClusterSingleton}} and commit that
separately.
Please do not merge the [PR|https://github.com/apache/lucene-solr/pull/1758]
in the curent form
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Fix For: master (9.0)
>
> Time Spent: 12.5h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17192974#comment-17192974
]
Ishan Chattopadhyaya commented on SOLR-14749:
-
I'd prefer if this were to be completely isolated away from core. Transitive
dependencies are supported today if all modules are in contrib and their jars
are in classpath. Anyway, let's refactor it after this is done. I can live with
this for now. +1 to dealing with triggers totally inside packages and not
making the concept first class. Thanks for your work on this and patience.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Time Spent: 9h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17192921#comment-17192921
]
Andrzej Bialecki commented on SOLR-14749:
-
[~ichattopadhyaya] as I explained in the comments that follow yours in the PR,
we actually need such coordination in Solr core. The concept of cluster
singleton instances is important, there are quite a few components currently
that work like this, and they all need this functionality. If Solr core doesn't
provide this then each plugin author will have to come up with their own hacky
and messy and buggy scheme.
If the package / plugin infra becomes more robust to support transitive
dependencies then we can extract the actual implementation of ClusterSingleton
management into its own plugin - but still, the interface needs to be in the
core, with a default implementation.
Again, this is not just for autoscaling - autoscaling will be just one user of
the ClusterSingleton concept.
I removed the Scheduler part from PR-1758. Do you still strongly object to the
current state of the PR, Ishan? If so, could you please outline technical
reasons for your objections?
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Time Spent: 9h
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17180444#comment-17180444
]
Ishan Chattopadhyaya commented on SOLR-14749:
-
I've mentioned it in this comment in the PR, but I'll repeat my intent here for
a broader audience.
https://github.com/apache/lucene-solr/pull/1758#discussion_r472263584
I *STRONGLY* oppose the idea of building a layer inside solr-core that will
help with the lifecycle management of plugins (the case in point here is for
the entire Solr cluster to have only one node that has a plugin in an active
state, i.e. "ClusterSingleton" as introduced in this PR). All such coordination
can be facilitated by a non-core plugin/package that autoscaling framework can
use. Such a coordination plugin can be used by other packages too, not just
autoscaling. However, if no such packages are in use, such a coordination layer
shouldn't even be visible to a user (either in form of running code or classes
loaded in his/her classpath). We should stop treating solr-core as a kitchen
sink.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Labels: clean-api
> Time Spent: 2h 50m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17179145#comment-17179145
]
Andrzej Bialecki commented on SOLR-14749:
-
See [https://github.com/apache/lucene-solr/pull/1758] for the initial draft.
Let's iterate.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177739#comment-17177739
]
Andrzej Bialecki commented on SOLR-14749:
-
{quote}If a plugin wishes to listen to a ZK node change, we should provide a
simple API to listen to that event.
{quote}
Autoscaling plugins should not need any knowledge of how Solr uses ZK for
distributed state mgmgt. This API should provide a clean separation from any ZK
internals and from how Solr implements stuff like live node tracking, or leader
election, or JSON files and their formats - so that the plugins can focus on
implementing their functionality and not on incidental aspects of ZK data.
We know very well that dealing with ZK is not straightforward under the best of
circumstances. Creating a new API that requires this knowledge is harmful in
long-term, even if it's easier short-term.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177290#comment-17177290
]
Ilan Ginzburg commented on SOLR-14749:
--
Not all plugins need to execute on one designated node. Most of the time the
constraint is have a single concurrent invocation.
For collection repair for example, we need a single concurrent invocation per
collection but it is likely desirable to be able to invoke repair for multiple
collections concurrently. And these repair triggers do not have to run in the
same JVM.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177221#comment-17177221
]
Ilan Ginzburg commented on SOLR-14749:
--
if there’s an abstraction layer between Solr and Autoscaling, then fixes to
Autoscaling can also be done independent of Solr.
If Autoscaling depends on the internals of Solr (like ZK) then changes to Solr
are more complex and/or break Autoscaling.
A plugin API should focus on the needs of the plugins, not on the current
implementation of Solr. It will be easier to use by plugin developers and not
lock us (Solr) into the current way we do things. Of course it has to reflect
Solr abstractions (node, collection, shard, replica) these likely change slower
than their implementations.
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
[jira] [Commented] (SOLR-14749) Provide a clean API for cluster-level event processing
[
https://issues.apache.org/jira/browse/SOLR-14749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177029#comment-17177029
]
Noble Paul commented on SOLR-14749:
---
In general my suggestions are as foloows.
* we should only have ability to have generic plugins in Solr
* We can have a new type of plugin that runs only on overseer
* Solr should not be aware of any events/triggers/actions
* If a plugin wishes to listen to a ZK node change, we should provide a simple
API to listen to that event.
This ensures that autoscaling is a self contained package and it all possible
bugfixes/improvements to autoscaling can be done independent of Solr
> Provide a clean API for cluster-level event processing
> --
>
> Key: SOLR-14749
> URL: https://issues.apache.org/jira/browse/SOLR-14749
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: AutoScaling
>Reporter: Andrzej Bialecki
>Assignee: Andrzej Bialecki
>Priority: Major
>
> This is a companion issue to SOLR-14613 and it aims at providing a clean,
> strongly typed API for the functionality formerly known as "triggers" - that
> is, a component for generating cluster-level events corresponding to changes
> in the cluster state, and a pluggable API for processing these events.
> The 8x triggers have been removed so this functionality is currently missing
> in 9.0. However, this functionality is crucial for implementing the automatic
> collection repair and re-balancing as the cluster state changes (nodes going
> down / up, becoming overloaded / unused / decommissioned, etc).
> For this reason we need this API and a default implementation of triggers
> that at least can perform automatic collection repair (maintaining the
> desired replication factor in presence of live node changes).
> As before, the actual changes to the collections will be executed using
> existing CollectionAdmin API, which in turn may use the placement plugins
> from SOLR-14613.
> h3. Division of responsibility
> * built-in Solr components (non-pluggable):
> ** cluster state monitoring and event generation,
> ** simple scheduler to periodically generate scheduled events
> * plugins:
> ** automatic collection repair on {{nodeLost}} events (provided by default)
> ** re-balancing of replicas (periodic or on {{nodeAdded}} events)
> ** reporting (eg. requesting additional node provisioning)
> ** scheduled maintenance (eg. removing inactive shards after split)
> h3. Other considerations
> These plugins (unlike the placement plugins) need to execute on one
> designated node in the cluster. Currently the easiest way to implement this
> is to run them on the Overseer leader node.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
-
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
