Website running on managed WP unreachable after latest automatic upgrade
-
@guyds thanks! I sent you a reply from support. But from your mail it seems it is not related to SQL query length as I initially thought but some bash scripting issue.
-
The issue here was the GROUP_CONCAT has a default length of 1024 or so. Setting session variable
SET group_concat_max_len=10000fixes the issue. -
@girish I had the same problem today following a Worpress update. I think the problem comes from the Civicrm plugin which has many tables. I restored with the last backup and disabled the automatic update, but that's not a solution.
-
@jeau Can you check if it's updating to v2.22.0-5 ? If you can clone from latest backup and update, I am happy to look into this if you can write to support@cloudron.io.
-
@jeau mm, per https://docs.civicrm.org/installation/en/latest/wordpress/, CiviCRM needs to be told about the prefix as well.
It's not entirely clear but it looks like we have to set
$db_prefix = 'wp_';in settings.php -
It seems civicrm does not use WP's
table_prefixat all. Well, I pushed a new package which simply ignore civicrm tables. -
@girish Ignoring the civicrm tables works.
However, I have the same kind of problem with an integration of MRBS in Wordpress for authentication. As the tables were properly renamed during the update, I simply set the ad hoc variable (
$db_tbl_prefix = "wp_mrbs_";) in the configuration file and it works.But I guess there are other use cases where renaming tables causes problems.
-
@jeau Good to know, thanks!
We have to do this migration since WordPress it says it doesn't support not having a table prefix. See:
