[JBoss JIRA] (TEIIDSB-144) Calling setTransactionIsolation on pg driver causes the connection to be closed
by Ramesh Reddy (Jira)
[ https://issues.redhat.com/browse/TEIIDSB-144?page=com.atlassian.jira.plug... ]
Ramesh Reddy commented on TEIIDSB-144:
--------------------------------------
Looks like if "tomcat-jdbc" is used, then URL property needs to be changed from "jdbc-url" to "url" this will have ripple effects all through teiid-syndesis and examples.
> Calling setTransactionIsolation on pg driver causes the connection to be closed
> -------------------------------------------------------------------------------
>
> Key: TEIIDSB-144
> URL: https://issues.redhat.com/browse/TEIIDSB-144
> Project: Teiid Spring Boot
> Issue Type: Bug
> Components: datasource
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Critical
> Fix For: 1.3.1
>
> Original Estimate: 4 hours
> Time Spent: 2 hours
> Remaining Estimate: 2 hours
>
> An attempt to set the isolation level on pg 42.2.8 resulted in an exception SQLSTATE 0a000 - unsupported - which causes the connection pool to mark the connection as broken:
> {code}
> 2019-11-25 23:07:50.125 WARN 43993 --- [ProcessorQueue6] com.zaxxer.hikari.pool.ProxyConnection : HikariPool-1 - Connection org.postgresql.jdbc.PgConnection@2ffebc7e marked as broken because of SQLSTATE(0A000), ErrorCode(0)
> org.postgresql.util.PSQLException:
> at org.postgresql.jdbc.PgConnection.setTransactionIsolation(PgConnection.java:851) ~[postgresql-42.2.8.jar:42.2.8]
> at com.zaxxer.hikari.pool.ProxyConnection.setTransactionIsolation(ProxyConnection.java:407) ~[HikariCP-3.2.0.jar:na]
> at com.zaxxer.hikari.pool.HikariProxyConnection.setTransactionIsolation(HikariProxyConnection.java) ~[HikariCP-3.2.0.jar:na]
> at org.teiid.translator.jdbc.JDBCBaseExecution.<init>(JDBCBaseExecution.java:72) ~[translator-jdbc-12.3.0.jar:12.3.0]
> {code}
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5876) MongoDB import name conflict issues
by Steven Hawkins (Jira)
Steven Hawkins created TEIID-5876:
-------------------------------------
Summary: MongoDB import name conflict issues
Key: TEIID-5876
URL: https://issues.redhat.com/browse/TEIID-5876
Project: Teiid
Issue Type: Bug
Components: Misc. Connectors
Reporter: Steven Hawkins
Assignee: Steven Hawkins
Fix For: 13.x
There are two issues with the import logic and naming:
1. There's an assumption that any table in the schema was created by the mongodb import. In general this is not true as other imports or ddl may have already run.
2. If a child construct has a field name that matches an existing table, the two entities will get intertwined in the metadata, which is invalid.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5875) MongoDB metadata import can create tables with no columns
by Steven Hawkins (Jira)
[ https://issues.redhat.com/browse/TEIID-5875?page=com.atlassian.jira.plugi... ]
Steven Hawkins resolved TEIID-5875.
-----------------------------------
Resolution: Done
Added a final check of the tables added, and removed any that are invalid. The name conflict issue will be handled separately.
> MongoDB metadata import can create tables with no columns
> ---------------------------------------------------------
>
> Key: TEIID-5875
> URL: https://issues.redhat.com/browse/TEIID-5875
> Project: Teiid
> Issue Type: Bug
> Components: Misc. Connectors
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Major
> Fix For: 13.1, 13.0.2
>
> Original Estimate: 3 hours
> Time Spent: 5 hours
> Remaining Estimate: 0 minutes
>
> The mongodb metadata import adds tables before determining what columns are available. In some circumstances this results in tables with no columns.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5875) MongoDB metadata import can create tables with no columns
by Steven Hawkins (Jira)
[ https://issues.redhat.com/browse/TEIID-5875?focusedWorklogId=12449710&pag... ]
Steven Hawkins logged work on TEIID-5875:
-----------------------------------------
Author: Steven Hawkins
Created on: 10/Jan/20 3:16 PM
Start Date: 10/Jan/20 3:15 PM
Worklog Time Spent: 5 hours
Work Description: It took a while to come up with a scenario under which there was no columns.
Issue Time Tracking
-------------------
Remaining Estimate: 0 minutes (was: 3 hours)
Time Spent: 5 hours
Worklog Id: (was: 12449710)
> MongoDB metadata import can create tables with no columns
> ---------------------------------------------------------
>
> Key: TEIID-5875
> URL: https://issues.redhat.com/browse/TEIID-5875
> Project: Teiid
> Issue Type: Bug
> Components: Misc. Connectors
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Major
> Fix For: 13.1, 13.0.2
>
> Original Estimate: 3 hours
> Time Spent: 5 hours
> Remaining Estimate: 0 minutes
>
> The mongodb metadata import adds tables before determining what columns are available. In some circumstances this results in tables with no columns.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5875) MongoDB metadata import can create tables with no columns
by Steven Hawkins (Jira)
[ https://issues.redhat.com/browse/TEIID-5875?page=com.atlassian.jira.plugi... ]
Steven Hawkins commented on TEIID-5875:
---------------------------------------
The simplest scenario in which this can happen is with a mongodb view that suppresses the _id field. That view is still reported as a collection and so it's easy to sample a row only contains a reference to another ojbect - in which case without an id and other columns a table with no columns is created.
> MongoDB metadata import can create tables with no columns
> ---------------------------------------------------------
>
> Key: TEIID-5875
> URL: https://issues.redhat.com/browse/TEIID-5875
> Project: Teiid
> Issue Type: Bug
> Components: Misc. Connectors
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Major
> Fix For: 13.1, 13.0.2
>
>
> The mongodb metadata import adds tables before determining what columns are available. In some circumstances this results in tables with no columns.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIIDSB-155) Infinispan configuration should support a transactional cache
by Steven Hawkins (Jira)
[ https://issues.redhat.com/browse/TEIIDSB-155?page=com.atlassian.jira.plug... ]
Steven Hawkins edited comment on TEIIDSB-155 at 1/10/20 11:38 AM:
------------------------------------------------------------------
The take-aways:
1. The underlying exception is misleading. It's not an authorization issue, it's that if the cache would be created using the default template it would not support transactions.
2. You have to use an xml configuration (or install a cache template) that specifies a transactional cache. It also must specify locking PESSIMISTIC as the default OPTIMISTIC requires versioning. I do not see any built-in cache templates that are transactional. This means that our connection code is effectively responsible for fully defining the cache - which does not seem ideal. That probably needs externalized in some way. Also this only seems to work against standalone. With the developer account against an instance in openshift we still are failing.
3. getCache will throw an exception in the transactional case rather than returning a null cache. The connection code needs to be updated accordingly.
was (Author: shawkins):
The take-aways:
1. The underlying exception is misleading. It's not an authorization issue, it's that if the cache would be created using the default template it would not support transactions.
2. You have to use an xml configuration (or install a cache template) that specifies a transactional cache. It also must specify locking PESSIMISTIC as the default OPTIMISTIC requires versioning. I do not see any built-in cache templates that are transactional. This means that our connection code is effectively responsible for fully defining the cache - which does not seem ideal. That probably needs externalized in some way.
3. getCache will throw an exception in the transactional case rather than returning a null cache. The connection code needs to be updated accordingly.
> Infinispan configuration should support a transactional cache
> -------------------------------------------------------------
>
> Key: TEIIDSB-155
> URL: https://issues.redhat.com/browse/TEIIDSB-155
> Project: Teiid Spring Boot
> Issue Type: Enhancement
> Components: datasource
> Reporter: Steven Hawkins
> Priority: Major
> Fix For: 1.3.1
>
>
> The configuration should allow for single phase commit and non-recoverable xa. This will need to change the transaction support on the translator, set the transaction config, and register a transactionManager lookup that can find the configured transaction manager.
> The infinispan operator does not provide a way to configure the transaction support of the cache, so this may not fully work yet - I'll test to see what happens both with the default cache and with trying to create one on demand.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5875) MongoDB metadata import can create tables with no columns
by Steven Hawkins (Jira)
[ https://issues.redhat.com/browse/TEIID-5875?page=com.atlassian.jira.plugi... ]
Steven Hawkins commented on TEIID-5875:
---------------------------------------
A related issue is that the logic will assume that child column names are distinct from collection names. That does not seem to be valid. It means if there is a collection called "table" and it has a column called table that contains a nested list/object it will be associated with the parent table.
> MongoDB metadata import can create tables with no columns
> ---------------------------------------------------------
>
> Key: TEIID-5875
> URL: https://issues.redhat.com/browse/TEIID-5875
> Project: Teiid
> Issue Type: Bug
> Components: Misc. Connectors
> Reporter: Steven Hawkins
> Assignee: Steven Hawkins
> Priority: Major
> Fix For: 13.1, 13.0.2
>
>
> The mongodb metadata import adds tables before determining what columns are available. In some circumstances this results in tables with no columns.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months
[JBoss JIRA] (TEIID-5875) MongoDB metadata import can create tables with no columns
by Steven Hawkins (Jira)
Steven Hawkins created TEIID-5875:
-------------------------------------
Summary: MongoDB metadata import can create tables with no columns
Key: TEIID-5875
URL: https://issues.redhat.com/browse/TEIID-5875
Project: Teiid
Issue Type: Bug
Components: Misc. Connectors
Reporter: Steven Hawkins
Assignee: Steven Hawkins
Fix For: 13.1, 13.0.2
The mongodb metadata import adds tables before determining what columns are available. In some circumstances this results in tables with no columns.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
5 years, 2 months