Skip to Content.
Sympa Menu

grouper-users - Re: [grouper-users] Grouper and MariaDB

Subject: Grouper Users - Open Discussion List

List archive

Re: [grouper-users] Grouper and MariaDB


Chronological Thread 
  • From: Julien Gribonvald <>
  • To:
  • Subject: Re: [grouper-users] Grouper and MariaDB
  • Date: Thu, 7 Feb 2019 11:45:57 +0100
  • Organization: GIP RECIA

Hi,

I would recommand that you watch on my configuration : https://github.com/GIP-RECIA/docker-mariadb/blob/master/.docker/mariadb-strict/conf/esco-strict-server.cnf

It's close to our production conf, it's optimized and it run pretty well with grouper 2.4, and also I would recommand the use of the option of one file per table, it saved us when a log table took to much space on hard drive, a truncate recreate the file and so reduced the used space, without that it saves all in one file and a truncate doesn't free space, so you will need to stop your service for maintenance, dump the database and import it (without logs) !

Best

- Julien

Le 06/02/2019 à 21:54, Richard Frovarp a écrit :
The part I left off was I'm on Ubuntu 18.04. I found this:

https://mariadb.com/kb/en/library/xtradbinnodb-storage-formats/

"Until MariaDB 10.2.1, the default was compact. Since MariaDB 10.2.2, the default is dynamic."

So I just used the MariaDB packages to go to 10.2 (which I think will match what we have on RHEL in production). That goes with the dynamic row format by default, and then everything worked without me changing a config file for MariaDB, or inserting dynamic on to the end off all of the create statements.

On 2/6/19 10:02 AM, Tim Darby wrote:
ROW_FORMAT=DYNAMIC fixed it for me, but you also need to make sure that the database parameter innodb_file_format is set to Barracuda. I think that requires a database restart.

Tim Darby
UA Identity Services | The University of Arizona |  | +1 520.626.3799


On Wed, Feb 6, 2019 at 7:56 AM Redman, Chad <> wrote:
From my reading, it does sound like that is the fix. It's an issue specific to InnoDB plus COMPACT or REDUNDANT row format. The limit for utf8mb4 would be 191 bytes, so there will probably be a lot more errors like that.  -Chad

-----Original Message-----
From: [mailto:] On Behalf Of Richard Frovarp
Sent: Wednesday, February 06, 2019 9:37 AM
To: Mailing List <>
Subject: [grouper-users] Grouper and MariaDB

How does one get Grouper 2.4 to work with MariaDB 10.1? I'm running into
the "Index column size too large. The maximum column size is 767 bytes."
error. Using utf8mb4.

I'm following the instructions from here:
https://spaces.at.internet2.edu/display/Grouper/Grouper+book+-+Setting+up+the+repository+database

I'm using the org.hibernate.dialect.MariaDB10Dialect dialect in
grouper.hibernate.properties.

I haven't gone through and fully inspected the generated file, but this
is generating the first error:


CREATE TABLE grouper_members
(
     id VARCHAR(40) NOT NULL,
     subject_id VARCHAR(255) NOT NULL,
     subject_source VARCHAR(255) NOT NULL,
     subject_type VARCHAR(255) NOT NULL,
     hibernate_version_number BIGINT,
     subject_identifier0 VARCHAR(255) NULL,
     sort_string0 VARCHAR(50) NULL,
     sort_string1 VARCHAR(50) NULL,
     sort_string2 VARCHAR(50) NULL,
     sort_string3 VARCHAR(50) NULL,
     sort_string4 VARCHAR(50) NULL,
     search_string0 VARCHAR(2048) NULL,
     search_string1 VARCHAR(2048) NULL,
     search_string2 VARCHAR(2048) NULL,
     search_string3 VARCHAR(2048) NULL,
     search_string4 VARCHAR(2048) NULL,
     name VARCHAR(2048) NULL,
     description VARCHAR(2048) NULL,
     context_id VARCHAR(40) NULL,
     PRIMARY KEY (id)
);

CREATE UNIQUE INDEX member_subjectsourcetype_idx ON grouper_members
(subject_id, subject_source, subject_type);

Is there a setting I need? I think adding ROW_FORMAT=DYNAMIC should fix
it, but wondering if there is an easier way. Or do I need a newer
MariaDB? Or should I be opening an issue?

Thanks,

Richard


--
Julien Gribonvald



Archive powered by MHonArc 2.6.19.

Top of Page