**Prevention of running 2 simultaneous updates**
Instead of using `RemoteMirror#update_status` and raise an error if
it's already running to prevent the same mirror being updated at the
same time we now use `Gitlab::ExclusiveLease` for that.
When we fail to obtain a lease in 3 tries, 30 seconds apart, we bail
and reschedule. We'll reschedule faster for the protected branches.
If the mirror already ran since it was scheduled, the job will be
skipped.
**Error handling: Remote side**
When an update fails because of a `Gitlab::Git::CommandError`, we
won't track this error in sentry, this could be on the remote side:
for example when branches have diverged.
In this case, we'll try 3 times scheduled 1 or 5 minutes apart.
In between, the mirror is marked as "to_retry", the error would be
visible to the user when they visit the settings page.
After 3 tries we'll mark the mirror as failed and notify the user.
We won't track this error in sentry, as it's not likely we can help
it.
The next event that would trigger a new refresh.
**Error handling: our side**
If an unexpected error occurs, we mark the mirror as failed, but we'd
still retry the job based on the regular sidekiq retries with
backoff. Same as we used to
The error would be reported in sentry, since its likely we need to do
something about it.
We call `Project#mark_stuck_remote_mirrors_as_failed!` from the
`Git::BaseHooksService`. So that gets called every time we push tags
or branches.
Before this would only mark started mirrors as failed if they had
been started 24 hours ago. A push would never take 24 hours,
especially not when we run it so often.
Lowering that threshold 1 hour should at least allow us to retry
broken mirrors more often on pushes.
The timeout for the initial push is set somewhat longer to accommodate
for pushing large repos. Both numbers are currently picked arbitrarily.
Retries in Sidekiq and in the remote mirror scheduler can cause repeated
attempts in quick succession if the sync fails. Each failure will then
send an e-mail to all project maintainers, which can spam users
unnecessarily.
Modify the logic to send one notification the first time the mirror
fails by setting `error_notification_sent` to `true` and reset the
flag after a successful sync.
Closes https://gitlab.com/gitlab-org/gitlab-ce/issues/56222
From https://gitlab.com/gitlab-org/gitlab-ce/issues/53515, we see the
backend appears to have inconsistent default values for this column:
* DB schema: false by default
* UI checkbox: false by default
* `RemoteMirror` model: true by default
This leads to unintended behavior where the boolean is activated if the
UI doesn't pass in a value for the checkbox.
Cleanup code, and refactor tests that still use Rugged. After this, there should
be no Rugged code that access the instance's repositories on non-test
environments. There is still some rugged code for other tasks like the
repository import task, but since it doesn't access any repository storage path
it can stay.
Remote mirrors only get created when the URL changes, However, during the GCP
migration, the remote mirror did not get created automatically. Plus, there's
no guarantee someone restoring a repository from backup would have this
remote. We now add the remote each time we attempt to fetch from the
repository.
This works because Gitaly doesn't throw up an exception or error if the
remote already exists:
https://gitlab.com/gitlab-org/gitaly/issues/1317
In the future, we should attempt to add if the remote doesn't exist:
https://gitlab.com/gitlab-org/gitaly/issues/1316Closes#50562
GitLab Bot
Grzegorz Bizon
Ash McKenzie
Lee Tickett
Bob Van Landuyt
Bob Van Landuyt
Luke Bennett
Thong Kuah
Stan Hu
John Jarvis
Francisco Javier López
Alejandro Rodríguez
Nick Thomas
Lin Jen-Shin
Jacob Vosmaer (GitLab)
Tiago Botelho