summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRémy Coutable <remy@rymai.me>2016-07-08 10:18:42 +0000
committerRémy Coutable <remy@rymai.me>2016-07-08 10:18:42 +0000
commit01d1be7bf1ffca28ce56dadeffcb426acfe8b0bf (patch)
tree1e84591d64898085ee6564b34f634d7d27c80a8d
parent2c650b6f30d2e76cf632d6ce5771859aed67ac48 (diff)
parent2c6fe72265d250e47c03f27dc274b59d3e7e93f5 (diff)
downloadgitlab-ce-01d1be7bf1ffca28ce56dadeffcb426acfe8b0bf.tar.gz
Merge branch 'fix/import-url-migration' into 'master'
import_url migration performance improvements Nullifying empty import_urls upfront so the number of projects with import_url not NULL decreases to 1/5. Also, now processing batches in blocks of 1000, with a threaded process - a bit experimental. See merge request !5149
-rw-r--r--db/migrate/20160620110927_fix_no_validatable_import_url.rb32
1 files changed, 26 insertions, 6 deletions
diff --git a/db/migrate/20160620110927_fix_no_validatable_import_url.rb b/db/migrate/20160620110927_fix_no_validatable_import_url.rb
index 82a616c62d9..a3f5073d511 100644
--- a/db/migrate/20160620110927_fix_no_validatable_import_url.rb
+++ b/db/migrate/20160620110927_fix_no_validatable_import_url.rb
@@ -11,7 +11,7 @@ class FixNoValidatableImportUrl < ActiveRecord::Migration
attr_reader :results, :query
- def initialize(batch_size: 100, query:)
+ def initialize(batch_size: 1000, query:)
@offset = 0
@batch_size = batch_size
@query = query
@@ -58,22 +58,38 @@ class FixNoValidatableImportUrl < ActiveRecord::Migration
return
end
+ say('Nullifying empty import URLs')
+
+ nullify_empty_urls
+
say('Cleaning up invalid import URLs... This may take a few minutes if we have a large number of imported projects.')
- invalid_import_url_project_ids.each { |project_id| cleanup_import_url(project_id) }
+ process_invalid_import_urls
end
- def invalid_import_url_project_ids
- ids = []
+ def process_invalid_import_urls
batches = SqlBatches.new(query: "SELECT id, import_url FROM projects WHERE import_url IS NOT NULL")
while batches.next?
+ project_ids = []
+
batches.results.each do |result|
- ids << result['id'] unless valid_url?(result['import_url'])
+ project_ids << result['id'] unless valid_url?(result['import_url'])
end
+
+ process_batch(project_ids)
end
- ids
+ end
+
+ def process_batch(project_ids)
+ Thread.new do
+ begin
+ project_ids.each { |project_id| cleanup_import_url(project_id) }
+ ensure
+ ActiveRecord::Base.connection.close
+ end
+ end.join
end
def valid_url?(url)
@@ -83,4 +99,8 @@ class FixNoValidatableImportUrl < ActiveRecord::Migration
def cleanup_import_url(project_id)
execute("UPDATE projects SET import_url = NULL WHERE id = #{project_id}")
end
+
+ def nullify_empty_urls
+ execute("UPDATE projects SET import_url = NULL WHERE import_url = ''")
+ end
end