The recommended Git type try Git v2.28. Minimal requisite type of Git v2.24 remains the same.
GitLab installations which have several web nodes needs to be improved to 13.1 before improving to 13.2 (and later) because of a splitting improvement in rail that lead to agreement problems.
2.0 remediates a message verification sidestep. After updating, if a number of your own users were all of a sudden experiencing 404 or 422 mistakes when finalizing in, or aˆ?blockedaˆ? information when using the command range, their accounts may have been un-confirmed. In this case, kindly keep these things see their mail for a re-confirmation hyperlink. For more information, read our topic of e-mail confirmation problem.
2.0 utilizes the btree_gist expansion for PostgreSQL. For installations with an externally handled PostgreSQL create, please remember to download the expansion manually before upgrading GitLab in the event the database consumer for GitLab isn’t a superuser. This is not necessary for installations using a GitLab managed PostgreSQL databases.
Y production (
- At the very least Git v2.24 (formerly, minimal called for adaptation is Git v2.22).
- Advised Git v2.26.
Troubles to accomplish this causes internal errors for the Gitaly solution in certain RPCs because of the utilization of the new –end-of-options Git flag.
Moreover, in 1.0, the version of Rails is improved from 6.0.3 to 6.0.3.1. The Rails improvement provided a change to CSRF token generation which will be not backwards-compatible – GitLab hosts using the latest rail variation generate CSRF tokens that are not identifiable by GitLab hosts using older Rails adaptation – that could trigger non-GET desires to fail for multi-node GitLab installations.
Therefore, if you use several Rails computers and especially updating from 13.0, all computers must very first feel upgraded to 13.1.Z before updating to 13.2.0 or later on:
But period cookie downgrades aren’t supported. So after upgrading to 12.2.0, any downgrades would lead to all sessions becoming invalidated and consumers are logged out.
Should you decide to improve from 12.0.Z to .Z , it’s important to do an intermediary improvement to 12.1.Z before improving to .Z to prevent issues like #215141.
In 12.0.0 we generated various databases connected modifications. These changes require that users very first upgrade to your current patch launch. After improved to .Z, customers can upgrade to 12.0.Z. Failure to take action may trigger database migrations not being applied, which may create software errors.
It’s also needed that you update to 12.0.Z before relocating to a future version of 12.Y.
Sample 1: you will be at this time utilizing GitLab .8, the latest plot production for .Z. It is possible to upgrade as always to 12.0.Z.
Instance 2: you may be currently making use of a form of GitLab 10.Y. To upgrade, first upgrade towards the finally 10.Y production (10.8.7) then your last 11.8). After enhanced to .8 you can easily securely update to 12.0.Z.
GitLab 13
Consumers have been closed in before upkeep setting is allowed will still be signed in. When the admin whom allowed repair setting manages to lose her program, they will be unable to disable Maintenance means through the UI. If that’s the case, you’ll disable upkeep function through the API or Rails r / spruzzo unit.
This insect is set in GitLab 14.5.0, and it is likely to be backported to GitLab 14.3 and 14.4.
Are you aware that artifacts, the GitLab athlete tries to upload them 3 x, and after that work sooner fails.
- ci_build_needs
4.0 include a back ground migration to move all leftover repositories in heritage storage to hashed storage. There are identified difficulties with this migration that are solved in 5.4 and soon after. Preferably, avoid 13.4.0 and update to 13.5.4 or maybe more alternatively. Remember that the migration may take quite a while to operate, dependent on what number of repositories should be moved. Make sure you be sure all history migrations have complete before upgrading more.