[ https://issues.apache.org/jira/browse/IGNITE-26346?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Denis Chudov updated IGNITE-26346: ---------------------------------- Description: The creation of zones without data nodes and alteration of zones removing all data nodes from them should be banned. For example: we have nodes A and B, having the node attribute {{{}attr{}}}: * A: attr=1 * B: attr=1 This command will fail: {{CREATE ZONE zone0 (NODES FILTER 'attr=2')}} This alteration will also fail: {{ALTER ZONE zone0 SET (NODES FILTER 'attr=2')}} The reasons for this: * recently we banned the creation of zones referring to storage profiles that don't exist in the cluster. We want to have the same approach with other things filtering the data nodes for distribution zones: they shouldn't filter out all nodes making the zone empty. * user mistakes and typos are the most common reasons for having empty data nodes; generally it just makes no sense. The only other reason may be running some setup script when some nodes in the cluster are not started yet, but it doesn't look like mandatory scenario to support. was: The alteration of zones removing all data nodes from them also should be banned. For example: we have nodes A and B, having the node attribute {{{}attr{}}}: * A: attr=1 * B: attr=1 This command will fail: {{CREATE ZONE zone0 (NODES FILTER 'attr=2')}} This alteration will also fail: {{ALTER ZONE zone0 SET (NODES FILTER 'attr=2')}} {{}} {{{}{}}}The reasons for this: * recently we banned the creation of zones referring to storage profiles that don't exist in the cluster. We want to have the same approach with other things filtering the data nodes for distribution zones: they shouldn't filter out all nodes making the zone empty. * user mistakes and typos are the most common reasons for having empty data nodes; generally it just makes no sense. The only other reason may be running some setup script when some nodes in the cluster are not started yet, but it doesn't look like mandatory scenario to support. > Ban the creation of distribution zones which don't have data nodes at the > moment of creation > -------------------------------------------------------------------------------------------- > > Key: IGNITE-26346 > URL: https://issues.apache.org/jira/browse/IGNITE-26346 > Project: Ignite > Issue Type: Improvement > Reporter: Denis Chudov > Assignee: Denis Chudov > Priority: Major > Labels: ignite-3 > > The creation of zones without data nodes and alteration of zones removing all > data nodes from them should be banned. > For example: we have nodes A and B, having the node attribute {{{}attr{}}}: * > A: attr=1 > * B: attr=1 > This command will fail: {{CREATE ZONE zone0 (NODES FILTER 'attr=2')}} > This alteration will also fail: {{ALTER ZONE zone0 SET (NODES FILTER > 'attr=2')}} > The reasons for this: * recently we banned the creation of zones referring to > storage profiles that don't exist in the cluster. We want to have the same > approach with other things filtering the data nodes for distribution zones: > they shouldn't filter out all nodes making the zone empty. > * user mistakes and typos are the most common reasons for having empty data > nodes; generally it just makes no sense. The only other reason may be running > some setup script when some nodes in the cluster are not started yet, but it > doesn't look like mandatory scenario to support. -- This message was sent by Atlassian Jira (v8.20.10#820010)