[Tickets #9835] Re: db schema update for h3->h4 converstion

bugs at horde.org bugs at horde.org
Thu May 19 21:24:24 UTC 2011


Ticket URL: http://bugs.horde.org/ticket/9835
  Ticket             | 9835
  Updated By         | dan at soleks.com
  Summary            | db schema update for h3->h4 converstion
  Queue              | Horde Base
  Version            | 4.0
  Type               | Bug
  State              | No Feedback
  Priority           | 2. Medium
  Milestone          |
  Patch              |
  Owners             |

dan at soleks.com (2011-05-19 21:24) wrote:

I did exactly what you described in your post - manually dropped  
indexes, then upgrade script was able to create new ones. Thanks that  
you confirm my theory.

> I see this problem.
> The reason it happens on my system, is that the primary key constraints have
> different names than those that the upgrade script is expecting. It  
> attempts to
> drop the primary key using an incorrect name which fails, then adds it back
> again. The add of course fails with the reported message.
> The postgres version is 8.4.2 on centos5, although the database was  
> originally
> created on a much earlier version.
> Running the following commands before the upgrade, allows the upgrade to
> run on my installation.
> alter table imp_sentmail drop constraint imp_primary_idx;
> alter table turba_shares drop constraint turba_shares_pkey_idx;
> alter table kronolith_shares drop constraint kronolith_shares_pkey_idx;
> alter table ingo_shares drop constraint ingo_shares_primary_idx;
> alter table ingo_rules drop constraint ingo_rules_primary_idx;
> alter table mnemo_shares drop constraint mnemo_shares_pkey_idx;
> alter table nag_shares drop constraint nag_shares_pkey_idx;
> alter table horde_groups drop constraint group_primary_idx;
> alter table horde_histories drop constraint history_primary_idx;
> alter table horde_perms drop constraint perms_primary_idx;
> alter table horde_vfs drop constraint vfs_primary_idx;
> There is then a problem with rampage_types which is resolved by running the
> failed upgrade again.

More information about the bugs mailing list