Clean up migration code. Defining migration custom class in only post migration file which requires it for each_batch

This commit is contained in:
Shinya Maeda 2018-06-07 14:18:00 +09:00
parent 5cfe73318b
commit 3780112298
2 changed files with 36 additions and 49 deletions

View File

@ -8,19 +8,32 @@ class MigrateLegacyArtifactsToJobArtifacts < ActiveRecord::Migration
disable_ddl_transaction!
class Build < ActiveRecord::Base
include EachBatch
self.table_name = 'ci_builds'
self.inheritance_column = :_type_disabled
scope :with_legacy_artifacts, -> { where("artifacts_file <> ''") }
scope :without_new_artifacts, -> do
where('NOT EXISTS (SELECT 1 FROM ci_job_artifacts WHERE (ci_builds.id = ci_job_artifacts.job_id) AND ci_job_artifacts.file_type = 1)')
end
end
def up
##
# We add a temporary index to `artifacts_file`. If we don't have the index,
# the first query (`SELECT .. WHERE .. ORDER BY id ASC LIMIT 1``) of `each_batch` will likely fail by statement timeout.
# We add a temporary index to the `ci_builds.artifacts_file` column.
# Without the index, the first query (`SELECT .. WHERE .. ORDER BY id ASC LIMIT 1``) of `each_batch` will likely fail by statement timeout.
# The following querires which will be executed in backgroun migrartions are fine without the index,
# because it's scanned by using `BETWEEN` clause (e.g. 'id BETWEEN 0 AND 2000') at the first place.
# because it's scanned by using `BETWEEN` clause (e.g. 'id BETWEEN 0 AND 2000') at the beginning and narrow down target rows.
unless index_exists_by_name?(:ci_builds, TMP_INDEX)
if Gitlab::Database.postgresql?
add_concurrent_index :ci_builds, :artifacts_file, where: "artifacts_file <> ''", name: TMP_INDEX
end
end
::Gitlab::BackgroundMigration::MigrateLegacyArtifacts::Build
MigrateLegacyArtifactsToJobArtifacts::Build
.with_legacy_artifacts.without_new_artifacts.tap do |relation|
queue_background_migration_jobs_by_range_at_intervals(relation,
MIGRATION,

View File

@ -5,39 +5,10 @@
module Gitlab
module BackgroundMigration
class MigrateLegacyArtifacts
class Build < ActiveRecord::Base
include EachBatch
self.table_name = 'ci_builds'
self.inheritance_column = :_type_disabled
scope :with_legacy_artifacts, -> { where("artifacts_file <> ''") }
scope :without_new_artifacts, -> do
where('NOT EXISTS (SELECT 1 FROM ci_job_artifacts WHERE (ci_builds.id = ci_job_artifacts.job_id) AND ci_job_artifacts.file_type = 1)')
end
end
class JobArtifact < ActiveRecord::Base
self.table_name = 'ci_job_artifacts'
LOCAL_STORE = 1 # Equavalant to ObjectStorage::Store::LOCAL
enum file_type: {
archive: 1,
metadata: 2,
trace: 3
}
##
# File location of the file
# legacy_path: File.join(model.created_at.utc.strftime('%Y_%m'), model.project_id.to_s, model.id.to_s)
# hashed_path: File.join(disk_hash[0..1], disk_hash[2..3], disk_hash, creation_date, model.job_id.to_s, model.id.to_s)
enum file_location: {
hashed_path: nil,
legacy_path: 1
}
end
FILE_LOCAL_STORE = 1 # equal to ObjectStorage::Store::LOCAL
ARCHIVE_FILE_TYPE = 1 # equal to Ci::JobArtifact.file_types['archive']
METADATA_FILE_TYPE = 2 # equal to Ci::JobArtifact.file_types['metadata']
LEGACY_PATH_FILE_LOCATION = 1 # equal to Ci::JobArtifact.file_location['legacy_path']
def perform(start_id, stop_id)
ActiveRecord::Base.transaction do
@ -65,20 +36,20 @@ module Gitlab
SELECT project_id,
id,
artifacts_expire_at,
#{MigrateLegacyArtifacts::JobArtifact.file_locations['legacy_path']},
#{LEGACY_PATH_FILE_LOCATION},
created_at,
created_at,
artifacts_file,
artifacts_size,
COALESCE(artifacts_file_store, #{JobArtifact::LOCAL_STORE}),
#{MigrateLegacyArtifacts::JobArtifact.file_types['archive']}
COALESCE(artifacts_file_store, #{FILE_LOCAL_STORE}),
#{ARCHIVE_FILE_TYPE}
FROM ci_builds
WHERE id BETWEEN #{start_id.to_i} AND #{stop_id.to_i}
AND artifacts_file <> ''
AND NOT EXISTS (
SELECT 1 FROM ci_job_artifacts
WHERE (ci_builds.id = ci_job_artifacts.job_id)
AND ci_job_artifacts.file_type = #{MigrateLegacyArtifacts::JobArtifact.file_types['archive']})
AND ci_job_artifacts.file_type = #{ARCHIVE_FILE_TYPE})
EOF
end
@ -98,29 +69,32 @@ module Gitlab
SELECT project_id,
id,
artifacts_expire_at,
#{MigrateLegacyArtifacts::JobArtifact.file_locations['legacy_path']},
#{LEGACY_PATH_FILE_LOCATION},
created_at,
created_at,
artifacts_metadata,
NULL,
COALESCE(artifacts_metadata_store, #{JobArtifact::LOCAL_STORE}),
#{MigrateLegacyArtifacts::JobArtifact.file_types['metadata']}
COALESCE(artifacts_metadata_store, #{FILE_LOCAL_STORE}),
#{METADATA_FILE_TYPE}
FROM ci_builds
WHERE id BETWEEN #{start_id.to_i} AND #{stop_id.to_i}
AND artifacts_file <> '' AND artifacts_metadata <> ''
AND NOT EXISTS (
SELECT 1 FROM ci_job_artifacts
WHERE (ci_builds.id = ci_job_artifacts.job_id)
AND ci_job_artifacts.file_type = #{MigrateLegacyArtifacts::JobArtifact.file_types['metadata']})
AND ci_job_artifacts.file_type = #{METADATA_FILE_TYPE})
EOF
end
def delete_legacy_artifacts(start_id, stop_id)
ActiveRecord::Base.connection.execute <<-EOF.strip_heredoc
UPDATE ci_builds SET
artifacts_file = NULL, artifacts_size = NULL, artifacts_file_store = NULL,
artifacts_metadata = NULL, artifacts_metadata_store = NULL
WHERE id BETWEEN #{start_id.to_i} AND #{stop_id.to_i}
UPDATE ci_builds
SET artifacts_file = NULL,
artifacts_file_store = NULL,
artifacts_size = NULL,
artifacts_metadata = NULL,
artifacts_metadata_store = NULL
WHERE id BETWEEN #{start_id.to_i} AND #{stop_id.to_i}
AND (artifacts_file <> '' OR artifacts_metadata <> '')
EOF
end