Rework retry strategy for remote mirrors

**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.

1 files changed, 43 insertions, 14 deletions

diff --git a/app/models/remote_mirror.rb b/app/models/remote_mirror.rb
index 6b5605f9999..c9ee0653d86 100644
--- a/app/models/remote_mirror.rb
+++ b/app/models/remote_mirror.rb
@@ -4,6 +4,8 @@ class RemoteMirror < ApplicationRecord
   include AfterCommitQueue
   include MirrorAuthentication
 
+  MAX_FIRST_RUNTIME = 3.hours
+  MAX_INCREMENTAL_RUNTIME = 1.hour
   PROTECTED_BACKOFF_DELAY   = 1.minute
   UNPROTECTED_BACKOFF_DELAY = 5.minutes
 
@@ -31,11 +33,18 @@ class RemoteMirror < ApplicationRecord
 
   scope :enabled, -> { where(enabled: true) }
   scope :started, -> { with_update_status(:started) }
-  scope :stuck,   -> { started.where('last_update_at < ? OR (last_update_at IS NULL AND updated_at < ?)', 1.hour.ago, 3.hours.ago) }
+
+  scope :stuck, -> do
+    started
+      .where('(last_update_started_at < ? AND last_update_at IS NOT NULL)',
+             MAX_INCREMENTAL_RUNTIME.ago)
+      .or(where('(last_update_started_at < ? AND last_update_at IS NULL)',
+                MAX_FIRST_RUNTIME.ago))
+  end
 
   state_machine :update_status, initial: :none do
     event :update_start do
-      transition [:none, :finished, :failed] => :started
+      transition any => :started
     end
 
     event :update_finish do
@@ -46,9 +55,14 @@ class RemoteMirror < ApplicationRecord
       transition started: :failed
     end
 
+    event :update_retry do
+      transition started: :to_retry
+    end
+
     state :started
     state :finished
     state :failed
+    state :to_retry
 
     after_transition any => :started do |remote_mirror, _|
       Gitlab::Metrics.add_event(:remote_mirrors_running)
@@ -138,16 +152,27 @@ class RemoteMirror < ApplicationRecord
   end
 
   def updated_since?(timestamp)
-    last_update_started_at && last_update_started_at > timestamp && !update_failed?
+    return false if failed?
+
+    last_update_started_at && last_update_started_at > timestamp
   end
 
   def mark_for_delete_if_blank_url
     mark_for_destruction if url.blank?
   end
 
-  def mark_as_failed(error_message)
-    update_column(:last_error, Gitlab::UrlSanitizer.sanitize(error_message))
-    update_fail
+  def update_error_message(error_message)
+    self.last_error = Gitlab::UrlSanitizer.sanitize(error_message)
+  end
+
+  def mark_for_retry!(error_message)
+    update_error_message(error_message)
+    update_retry!
+  end
+
+  def mark_as_failed!(error_message)
+    update_error_message(error_message)
+    update_fail!
   end
 
   def url=(value)
@@ -190,6 +215,18 @@ class RemoteMirror < ApplicationRecord
     update_column(:error_notification_sent, true)
   end
 
+  def backoff_delay
+    if self.only_protected_branches
+      PROTECTED_BACKOFF_DELAY
+    else
+      UNPROTECTED_BACKOFF_DELAY
+    end
+  end
+
+  def max_runtime
+    last_update_at.present? ? MAX_INCREMENTAL_RUNTIME : MAX_FIRST_RUNTIME
+  end
+
   private
 
   def store_credentials
@@ -219,14 +256,6 @@ class RemoteMirror < ApplicationRecord
     self.last_update_started_at >= Time.now - backoff_delay
   end
 
-  def backoff_delay
-    if self.only_protected_branches
-      PROTECTED_BACKOFF_DELAY
-    else
-      UNPROTECTED_BACKOFF_DELAY
-    end
-  end
-
   def reset_fields
     update_columns(
       last_error: nil,