ERROR after upgrade: Unknown property actual type ordinal nu

[steve_tekell]

after upgrading from 2.1 to 3c I am getting a seemingly endless stream of these errors

13:26:33,577  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: -1
13:26:33,577  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: null
13:26:33,636  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: -1
13:26:33,636  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: -1
13:26:33,636  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: null
13:26:33,636  ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: null
…
‍‍‍‍‍‍‍‍

What are these referring to?
How do I fix?

[steve_tekell]

I have made some progess on the source of the problem
- node_properties originally of actual type MLTEXT and persisted type DB_ATTRIBUTE
see issue:
https://issues.alfresco.com/jira/browse/ALFCOM-2274

I am not sure if I can delete these rows now or if the original data should be rescued into something useful.

[digitech]

We have the error same messages.

Upgrade from 2.1 -> 3.0C impossible

we migrate from 2.1 to 2.9B TO labs 3.0C

   Config:  Win2003 as Vmware virtual machine  Mysql 5

Install standard, no personalisation

First af all: we are french and the Labs 3.0C apparently don't support accentuated characters ??
We had to modify several spaces.
Is there a mean to have accentuated caracters?

The web interface works fine

but
we have always such messages:
13:26:33,636 ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: -1
13:26:33,636 ERROR [repo.domain.NodePropertyValue] Unknown property actual type ordinal number: null
Where is the problem , how to correct it ?

The CIFS works but when a client connect : the CPU rise to 100% and stay at this level  ( java.exe 99%)
the server is blocked.
Where is the problem, the install is standard.??

We tried a Full re-index that aborted.
We re-install the index after migration?
Is it normal ?


Thanks for your help and y apologise for my english.

[steve_tekell]

Where is the problem , how to correct it ?

I put some notes here on where I think the problem is:
https://issues.alfresco.com/jira/browse/ALFCOM-2274

The SQL script, upgrade-from-2.1.sql, results in bad records in alf_node_properties
they have actual_type_n = -1 and their original value which was stored in the attribute_value column in the old alf_node_properties table has been lost.

The resolution will be probably be one of two things.
1) If the alfresco app no longer needs these lost values, we can just delete the bad rows and the error messages will stop.
2) If the alfresco app needs these lost values, then it gets tricky.  The upgrade script could be fixed and then you would need to revert and upgrade anew.  If reverting to the old database isn't an option, you'd have a find some way to restore the old table somewhere and do something to get these values and convert them into the new table correctly.

I hope to hear from Alfresco that it's #1.