Add support for s3 GLACIER storage class, when GET return error, th…

…en, it requires 5 minutes per key and restore could be slow. Use `GLACIER_IR`, it looks more robust, fix #614

ChangeLog.md

@@ -6,6 +6,7 @@ IMPROVEMENTS
 - Backup/Restore RBAC related objects from Zookeeper via direct connection to zookeeper/keeper, fix [604](https://github.com/Altinity/clickhouse-backup/issues/604)
 - Add `SHARDED_OPERATION_MODE` option, to easy create backup for sharded cluster, available values `none` (no sharding), `table` (table granularity), `database` (database granularity), `first-replica` (on the lexicographically sorted first active replica), thanks @mskwon, fix [639](https://github.com/Altinity/clickhouse-backup/issues/639), fix [648](https://github.com/Altinity/clickhouse-backup/pull/648)
 - Add support for `compression_format: none` for upload and download backups created with `--rbac` / `--rbac-only` or `--configs` / `--configs-only` options, fix [713](https://github.com/Altinity/clickhouse-backup/issues/713)
+- Add support for s3 `GLACIER` storage class, when GET return error, then, it requires 5 minutes per key and restore could be slow. Use `GLACIER_IR`, it looks more robust, fix [614](https://github.com/Altinity/clickhouse-backup/issues/614)  
 
 BUG FIXES
 - fix possible create backup failures during UNFREEZE not exists tables, affected 2.2.7+ version, fix [704](https://github.com/Altinity/clickhouse-backup/issues/704)

pkg/storage/s3.go

@@ -144,7 +144,7 @@ func (s *S3) Connect(ctx context.Context) error {
 
 	if s.Config.Debug {
 		awsConfig.Logger = newS3Logger(s.Log)
-		awsConfig.ClientLogMode = aws.LogRetries | aws.LogRequest | aws.LogResponse
+		awsConfig.ClientLogMode = aws.LogRetries | aws.LogRequest | aws.LogResponseWithBody
 	}
 
 	httpTransport := http.DefaultTransport
@@ -630,7 +630,8 @@ func (s *S3) restoreObject(ctx context.Context, key string) error {
 		}
 
 		if res.Restore != nil && *res.Restore == "ongoing-request=\"true\"" {
-			s.Log.Debugf("%s still not restored, will wait %d seconds", key, i*5)
+			i += 1
+			s.Log.Warnf("%s still not restored, will wait %d seconds", key, i*5)
 			time.Sleep(time.Duration(i*5) * time.Second)
 		} else {
 			return nil

test/integration/config-s3-fips.yml

@@ -35,7 +35,6 @@ s3:
   compression_format: tar
   allow_multipart_download: true
   concurrency: 3
-  storage_class: GLACIER
 api:
   listen: :7171
   create_integration_tables: true

test/integration/config-s3-glacier.yml

@@ -0,0 +1,44 @@
+general:
+  disable_progress_bar: true
+  remote_storage: s3
+  upload_concurrency: 4
+  download_concurrency: 4
+  skip_tables:
+    - " system.*"
+    - "INFORMATION_SCHEMA.*"
+    - "information_schema.*"
+    - "_temporary_and_external_tables.*"
+  restore_schema_on_cluster: "{cluster}"
+clickhouse:
+  host: clickhouse
+  port: 9440
+  username: backup
+  password: meow=& 123?*%# МЯУ
+  secure: true
+  skip_verify: true
+  sync_replicated_tables: true
+  timeout: 1h
+  restart_command: bash -c 'echo "FAKE RESTART"'
+  backup_mutations: true
+# secrets for `FISP` will provide from `.env` or from GitHub actions secrets
+s3:
+  access_key: ${QA_AWS_ACCESS_KEY}
+  secret_key: ${QA_AWS_SECRET_KEY}
+  bucket: ${QA_AWS_BUCKET}
+#  endpoint: https://${QA_AWS_BUCKET}.s3-fips.${QA_AWS_REGION}.amazonaws.com/
+  region: ${QA_AWS_REGION}
+  acl: private
+  force_path_style: false
+  path: backup/{cluster}/{shard}
+  object_disk_path: object_disks/{cluster}/{shard}
+  disable_ssl: false
+  compression_format: tar
+  allow_multipart_download: false
+  concurrency: 4
+  # storage_class: GLACIER, 6000 seconds test execution
+  storage_class: GLACIER_IR
+api:
+  listen: :7171
+  create_integration_tables: true
+  integration_tables_host: "localhost"
+  allow_parallel: false

test/integration/integration_test.go

@@ -952,7 +952,7 @@ func TestFIPS(t *testing.T) {
 		r.NoError(dockerExec("clickhouse", "bash", "-xce", "openssl req -subj \"/CN=localhost\" -addext \"subjectAltName = DNS:localhost,DNS:*.cluster.local\" -new -key /etc/clickhouse-backup/server-key.pem -out /etc/clickhouse-backup/server-req.csr"))
 		r.NoError(dockerExec("clickhouse", "bash", "-xce", "openssl x509 -req -days 365000 -extensions SAN -extfile <(printf \"\\n[SAN]\\nsubjectAltName=DNS:localhost,DNS:*.cluster.local\") -in /etc/clickhouse-backup/server-req.csr -out /etc/clickhouse-backup/server-cert.pem -CA /etc/clickhouse-backup/ca-cert.pem -CAkey /etc/clickhouse-backup/ca-key.pem -CAcreateserial"))
 	}
-	r.NoError(dockerExec("clickhouse", "bash", "-c", "cat /etc/clickhouse-backup/config.yml.fips-template | envsubst > /etc/clickhouse-backup/config.yml"))
+	r.NoError(dockerExec("clickhouse", "bash", "-xec", "cat /etc/clickhouse-backup/config.yml.fips-template | envsubst > /etc/clickhouse-backup/config.yml"))
 
 	generateCerts("rsa", "4096", "")
 	createSQL := "CREATE TABLE default.fips_table (v UInt64) ENGINE=MergeTree() ORDER BY tuple()"
@@ -1158,6 +1158,20 @@ func TestDoRestoreConfigs(t *testing.T) {
 	ch.chbackend.Close()
 }
 
+func TestIntegrationS3Glacier(t *testing.T) {
+	if isTestShouldSkip("GLACIER_TESTS") {
+		t.Skip("Skipping GLACIER integration tests...")
+		return
+	}
+	r := require.New(t)
+	r.NoError(dockerCP("config-s3-glacier.yml", "clickhouse-backup:/etc/clickhouse-backup/config.yml.s3glacier-template"))
+	installDebIfNotExists(r, "clickhouse-backup", "curl", "gettext-base", "bsdmainutils", "dnsutils", "git", "ca-certificates")
+	r.NoError(dockerExec("clickhouse-backup", "bash", "-xec", "cat /etc/clickhouse-backup/config.yml.s3glacier-template | envsubst > /etc/clickhouse-backup/config.yml"))
+	dockerExecTimeout = 60 * time.Minute
+	runMainIntegrationScenario(t, "GLACIER")
+	dockerExecTimeout = 3 * time.Minute
+}
+
 func TestIntegrationS3(t *testing.T) {
 	r := require.New(t)
 	r.NoError(dockerCP("config-s3.yml", "clickhouse-backup:/etc/clickhouse-backup/config.yml"))
@@ -2412,6 +2426,8 @@ func (ch *TestClickHouse) queryWithNoError(r *require.Assertions, query string,
 	r.NoError(err)
 }
 
+var dockerExecTimeout = 180 * time.Second
+
 func dockerExec(container string, cmd ...string) error {
 	out, err := dockerExecOut(container, cmd...)
 	log.Info(out)
@@ -2421,7 +2437,7 @@ func dockerExec(container string, cmd ...string) error {
 func dockerExecOut(container string, cmd ...string) (string, error) {
 	dcmd := []string{"exec", container}
 	dcmd = append(dcmd, cmd...)
-	return utils.ExecCmdOut(context.Background(), 180*time.Second, "docker", dcmd...)
+	return utils.ExecCmdOut(context.Background(), dockerExecTimeout, "docker", dcmd...)
 }
 
 func dockerCP(src, dst string) error {
@@ -2490,7 +2506,7 @@ func installDebIfNotExists(r *require.Assertions, container string, pkgs ...stri
 		container,
 		"bash", "-xec",
 		fmt.Sprintf(
-			"export DEBIAN_FRONTEND=noniteractive; if [[ '%d' != $(dpkg -l | grep -c -E \"%s\" ) ]]; then rm -fv /etc/apt/sources.list.d/clickhouse.list; find /etc/apt/ -type f -exec sed -i 's/ru.archive.ubuntu.com/archive.ubuntu.com/g' {} +; apt-get -y update; apt-get install --no-install-recommends -y %s; fi",
+			"export DEBIAN_FRONTEND=noniteractive; if [[ '%d' != $(dpkg -l | grep -c -E \"%s\" ) ]]; then rm -fv /etc/apt/sources.list.d/clickhouse.list; find /etc/apt/ -type f -name *.list -exec sed -i 's/ru.archive.ubuntu.com/archive.ubuntu.com/g' {} +; apt-get -y update; apt-get install --no-install-recommends -y %s; fi",
 			len(pkgs), "^ii\\s+"+strings.Join(pkgs, "|^ii\\s+"), strings.Join(pkgs, " "),
 		),
 	))

test/integration/run.sh

@@ -23,6 +23,8 @@ else
   export GCS_TESTS=${GCS_TESTS:-}
 fi
 
+export GLACIER_TESTS=${GLACIER_TESTS:-}
+
 export AZURE_TESTS=${AZURE_TESTS:-1}
 export RUN_ADVANCED_TESTS=${RUN_ADVANCED_TESTS:-1}
 export S3_DEBUG=${S3_DEBUG:-false}
@@ -44,5 +46,5 @@ make clean build-race-docker build-race-fips-docker
 docker-compose -f ${CUR_DIR}/${COMPOSE_FILE} up -d
 docker-compose -f ${CUR_DIR}/${COMPOSE_FILE} exec minio mc alias list
 
-go test -timeout 30m -failfast -tags=integration -run "${RUN_TESTS:-.+}" -v ${CUR_DIR}/integration_test.go
+go test -timeout ${TESTS_TIMEOUT:-30m} -failfast -tags=integration -run "${RUN_TESTS:-.+}" -v ${CUR_DIR}/integration_test.go
 go tool covdata textfmt -i "${CUR_DIR}/_coverage_/" -o "${CUR_DIR}/_coverage_/coverage.out"