Backup Datastore using the Admin REST API

Add server end points to backup Datastore using managed-export mechanism.
A cron job is defined in Alpha to run daily exports using this implementation.

Existing backup is left running. The new backups are saved to a new set of
locations:
- GCS bucket: gs://PROJECT-ID-datastore-backups
- Big Query data set: datastore_backups
- Big Query latest back up view name: latest_datastore_backup
Also, the names of Bigquery tables now use the export timestamp
assigned by Datastore. E.g., 2018_12_05T23_56_18_50532_ContactResource,

After the new import mechanism is implemented and the back-restore flow is
tested, we will stop the existing backup runs and deploy the new
implementation to all environments.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=224932957
This commit is contained in:
weiminyu 2018-12-10 20:26:53 -08:00 committed by jianglai
parent ea154a8378
commit 9c706e79fd
27 changed files with 1179 additions and 37 deletions

View file

@ -64,6 +64,8 @@ public class LoadSnapshotAction implements Runnable {
static final String SNAPSHOTS_DATASET = "snapshots";
static final String LATEST_SNAPSHOT_VIEW_NAME = "latest_datastore_export";
/** Servlet-specific details needed for enqueuing tasks against itself. */
static final String QUEUE = "export-snapshot"; // See queue.xml.
static final String PATH = "/_dr/task/loadSnapshot"; // See web.xml.
@ -131,7 +133,7 @@ public class LoadSnapshotAction implements Runnable {
// well-known view in BigQuery to point at the newly loaded snapshot table for this kind.
bigqueryPollEnqueuer.enqueuePollTask(
jobRef,
createViewUpdateTask(SNAPSHOTS_DATASET, tableId, kindName),
createViewUpdateTask(SNAPSHOTS_DATASET, tableId, kindName, LATEST_SNAPSHOT_VIEW_NAME),
getQueue(UpdateSnapshotViewAction.QUEUE));
builder.append(String.format(" - %s:%s\n", projectId, jobId));