Jira Import from H2DB to Oracle causes ORA-00972: identifier is too long

[Roman Kharytonov]

During upgrade from 6.14 to more recent version the Jira has been upgraded first to 7.0.0 as suggested by official guides. Old version and intermediary version (7.0.0) are both running on in-memory DB. The target version is using Oracle 11g.

Import from 6.14 to 7.0.0 went fine. However, during import of backup from 7.0.0 to target Jira version this error occurs:

com.atlassian.activeobjects.spi.ActiveObjectsImportExportException: There was an error during import/export with plugin JIRA DVCS Connector Plugin(com.atlassian.jira.plugins.jira-bitbucket-connector-plugin) #4.0.14 (table AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12):Error executing update for SQL statement 'ALTER TABLE "AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12" DISABLE ALL TRIGGERS'
 at com.atlassian.activeobjects.backup.ImportExportErrorServiceImpl.newImportExportSqlException(ImportExportErrorServiceImpl.java:26)
......
Caused by: java.sql.SQLSyntaxErrorException: ORA-00972: identifier is too long

The identifier AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12 is 40 characters long which is above Oracle 11g's limit of 30 characters for identifiers. 

Moreover, in entities.xml DVCS plugin explicitly references "AO_E8B6CC_REPOSITORY_MAPPING" table. Where that extra comes from?

[Roman Kharytonov · Accepted solution]

Ok, so there  are few possible workarounds:

1) AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12 is created during upgrade to 7.0.0 but not used afterwards and can be safely dropped from schema as long as AO_E8B6CC_REPOSITORY_MAPPING_COPY is present. No idea why this gets created in first place because old Jira's schema doesn't have _COPY_ table

2) Unzip export file and rename all mentions of AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12 to AO_E8B6CC_REPOSITORY_MAPPING_COPY_X

sed -i -e 's/AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12/AO_E8B6CC_REPOSITORY_MAPPING_X/g' *.xml

3) Manually delete <table> and <data> sercions for AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12  from activeobjects.xml (won't work if file is too big)


Another issue is that after import 90% complete it throws "java.sql.SQLSyntaxErrorException: ORA-00904: "APPROVAL_STATE": invalid identifier" error. That's normal as such object is created in later upgrade stages. Once you get this exception simply restart Jira and let it run for a while (1-2 hours). Logs might look stale at that time but don't worry, upgrade is happening.

[Rambabu Patina _Appfire_]

Hi @Roman Kharytonov, You may follow the workaround here to resolve the issue.

[Tom Lister]

good to know ;-)

[Roman Kharytonov]

@Rambabu Patina _Appfire_ it's already defined as in mentioned KB 

For some reason Jira Backup process produces two tables:

AO_E8B6CC_REPOSITORY_MAPPING

and 

AO_E8B6CC_REPOSITORY_MAPPING_COPY_331_12

[Tom Lister]

Hi @Roman Kharytonov

Not an answer, but I can see it's to do with the JIRA DVCS Connector Plugin. Do you have latest version installed.

This is a case of the plugin not keeping up with restrictions on different platforms.

On the plugin page they don't seem to have released for a while. I think you will need to contact vendor if you can or support. May need to find a new plugin for CVS.

In the sort term you could try disabling that plugin and re-importing.

[Roman Kharytonov]

Hi @Tom Lister  

com.atlassian.jira.plugins.jira-bitbucket-connector-plugin cannot be uninstalled or disabled as this is core component of Jira

I'll trying to remove whole rows related to _COPY_ table in activeobjects.xml but it's not a quick process