[
https://issues.apache.org/jira/browse/IGNITE-23743?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kirill Gusakov updated IGNITE-23743:
------------------------------------
Description:
Currently it is possible to create a zone with invalid storage profile, but a
user receives no sign of error.
An error happens only when a user tries to interact with a table in the zone,
for example insert values. The error is Unknown error Failed to get the primary
replica [tablePartitionId=23_part_0, awaitTimestamp=HybridTimestamp
[physical=2024-11-19 13:12:44:718 +0000, logical=0,
composite=113509791479758848]].
{code:java}
sql-cli> CREATE ZONE IF NOT EXISTS exampleZone1 WITH STORAGE_PROFILES='defaul'
Updated 0 rows.
sql-cli> CREATE TABLE IF NOT EXISTS Person4 ( id int primary key, city_id int,
name varchar, age int, company varch ar) ZONE exampleZone1 Updated 0 rows.
sql-cli> INSERT INTO Person4 VALUES (1,2,'carl', 32, 'clover') Unknown error
Failed to get the primary replica [tablePartitionId=23_part_0,
awaitTimestamp=HybridTimestamp [physical=2024-11-19 13:12:44:718 +0000,
logical=0, composite=113509791479758848]] {code}
This should be changed somehow. The current behavior most likely appeared as a
consequence of https://issues.apache.org/jira/browse/IGNITE-23128.
*UPDATE*
In general we will have the same behaviour in all cases, when according to the
current zone filters (general filter and storage profile filter) we have no (or
not enough) available data nodes. We must:
* Change the type of public exception to PartitionIsNotAvailable
* Add the full zone description in the human readable manner to the message. It
will help to understand the possible issues with the zone filters (storage
profile is the filter also) on the user side.
* If the root cause - no alive majority for the partition - add the message
about it and the list of available data nodes.
* If the root cause - no alive data nodes with the needed attributes - add the
message about the empty data nodes.
* Print the list of available cluster nodes with the nodes' attributes. It will
help to find the issues with the node attributes.
was:
Currently it is possible to create a zone with invalid storage profile, but a
user receives no sign of error.
An error happens only when a user tries to interact with a table in the zone,
for example insert values. The error is Unknown error Failed to get the primary
replica [tablePartitionId=23_part_0, awaitTimestamp=HybridTimestamp
[physical=2024-11-19 13:12:44:718 +0000, logical=0,
composite=113509791479758848]].
{code:java}
sql-cli> CREATE ZONE IF NOT EXISTS exampleZone1 WITH STORAGE_PROFILES='defaul'
Updated 0 rows.
sql-cli> CREATE TABLE IF NOT EXISTS Person4 ( id int primary key, city_id int,
name varchar, age int, company varch ar) ZONE exampleZone1 Updated 0 rows.
sql-cli> INSERT INTO Person4 VALUES (1,2,'carl', 32, 'clover') Unknown error
Failed to get the primary replica [tablePartitionId=23_part_0,
awaitTimestamp=HybridTimestamp [physical=2024-11-19 13:12:44:718 +0000,
logical=0, composite=113509791479758848]] {code}
This should be changed somehow. The current behavior most likely appeared as a
consequence of https://issues.apache.org/jira/browse/IGNITE-23128.
> No errors returned when users create a distribution zone with an invalid
> storage profile
> ----------------------------------------------------------------------------------------
>
> Key: IGNITE-23743
> URL: https://issues.apache.org/jira/browse/IGNITE-23743
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladimir Pligin
> Priority: Major
> Labels: ignite-3
>
> Currently it is possible to create a zone with invalid storage profile, but a
> user receives no sign of error.
> An error happens only when a user tries to interact with a table in the zone,
> for example insert values. The error is Unknown error Failed to get the
> primary replica [tablePartitionId=23_part_0, awaitTimestamp=HybridTimestamp
> [physical=2024-11-19 13:12:44:718 +0000, logical=0,
> composite=113509791479758848]].
> {code:java}
> sql-cli> CREATE ZONE IF NOT EXISTS exampleZone1 WITH
> STORAGE_PROFILES='defaul' Updated 0 rows.
> sql-cli> CREATE TABLE IF NOT EXISTS Person4 ( id int primary key, city_id
> int, name varchar, age int, company varch ar) ZONE exampleZone1 Updated 0
> rows.
> sql-cli> INSERT INTO Person4 VALUES (1,2,'carl', 32, 'clover') Unknown error
> Failed to get the primary replica [tablePartitionId=23_part_0,
> awaitTimestamp=HybridTimestamp [physical=2024-11-19 13:12:44:718 +0000,
> logical=0, composite=113509791479758848]] {code}
> This should be changed somehow. The current behavior most likely appeared as
> a consequence of https://issues.apache.org/jira/browse/IGNITE-23128.
>
> *UPDATE*
> In general we will have the same behaviour in all cases, when according to
> the current zone filters (general filter and storage profile filter) we have
> no (or not enough) available data nodes. We must:
> * Change the type of public exception to PartitionIsNotAvailable
> * Add the full zone description in the human readable manner to the message.
> It will help to understand the possible issues with the zone filters (storage
> profile is the filter also) on the user side.
> * If the root cause - no alive majority for the partition - add the message
> about it and the list of available data nodes.
> * If the root cause - no alive data nodes with the needed attributes - add
> the message about the empty data nodes.
> * Print the list of available cluster nodes with the nodes' attributes. It
> will help to find the issues with the node attributes.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
