Use run ID instead of date prefix for resource naming.

batch-setup/batch.py

@@ -189,7 +189,7 @@ def s3_policy(bucket, date_prefix, allow_write=False):
 def find_or_create_s3_policy(iam, bucket_name, bucket, date_prefix,
                              allow_write=False):
     """
-    Finds a policy in the planet_date environment to access the given bucket,
+    Finds a policy in the run's environment to access the given bucket,
     or creates one.
 
     Returns the policy's ARN.
@@ -216,9 +216,9 @@ def kebab_to_camel(name):
     return "".join(map(lambda s: s.capitalize(), name.split('-')))
 
 
-def ensure_job_role_arn(iam, planet_date, name, buckets, date_prefixes):
+def ensure_job_role_arn(iam, run_id, name, buckets, date_prefixes):
     role_name = kebab_to_camel(
-        "batch-%s-%s" % (name, planet_date.strftime("%y%m%d")))
+        "batch-%s-%s" % (name, run_id))
 
     arn = None
     try:
@@ -259,7 +259,7 @@ def ensure_job_role_arn(iam, planet_date, name, buckets, date_prefixes):
 def create_role(iam, image_name, role_name, buckets, date_prefixes):
 
     """
-    Create a role with the given role_name for the image in the planet_date
+    Create a role with the given role_name for the image in the run's
     environment.
     """
 
@@ -317,7 +317,7 @@ def allow_s3_write(self, name, bucket, date_prefix):
 
 
 def make_job_definitions(
-        iam, planet_date, region, repo_urls, databases, buckets,
+        iam, run_id, region, repo_urls, databases, buckets,
         db_password, memory, vcpus, retry_attempts, date_prefixes,
         check_metatile_exists):
 
@@ -327,13 +327,13 @@ def make_job_definitions(
     definition_names = {}
     for name, image in repo_urls.items():
         job_role_arn = ensure_job_role_arn(
-            iam, planet_date, name, buckets, date_prefixes)
+            iam, run_id, name, buckets, date_prefixes)
         memory_value = memory[name] if isinstance(memory, dict) else memory
         vcpus_value = vcpus[name] if isinstance(vcpus, dict) else vcpus
         retry_value = retry_attempts[name] \
             if isinstance(retry_attempts, dict) else retry_attempts
 
-        job_name = "%s-%s" % (name, planet_date.strftime("%y%m%d"))
+        job_name = "%s-%s" % (name, run_id)
         definition = {
             'name': job_name,
             'job-role-arn': job_role_arn,
@@ -424,7 +424,7 @@ def run_go(cmd, *args, **kwargs):
             stdout.close()
 
 
-def create_job_definitions(planet_date, region, repo_urls, databases, buckets,
+def create_job_definitions(run_id, region, repo_urls, databases, buckets,
                            db_password, memory=1024, vcpus=1,
                            retry_attempts=5, date_prefix=None,
                            meta_date_prefix=None, check_metatile_exists=False):
@@ -437,7 +437,7 @@ def create_job_definitions(planet_date, region, repo_urls, databases, buckets,
     not they will be passed directly to AWS.
 
     If date_prefix is left as None (the default) it will be generated from
-    planet_date automatically.
+    run_id automatically.
 
     If meta_date_prefix is specified, a different date prefix will be used in
     the metatile and missing buckets.
@@ -448,15 +448,15 @@ def create_job_definitions(planet_date, region, repo_urls, databases, buckets,
     iam = boto3.client('iam')
 
     if date_prefix is None:
-        date_prefix = planet_date.strftime('%Y%m%d')
+        date_prefix = run_id
 
     if meta_date_prefix is None:
         meta_date_prefix = date_prefix
 
     date_prefixes = Buckets(date_prefix, meta_date_prefix, meta_date_prefix)
 
     job_definitions, definition_names = make_job_definitions(
-        iam, planet_date, region, repo_urls, databases, buckets,
+        iam, run_id, region, repo_urls, databases, buckets,
         db_password, memory, vcpus, retry_attempts, date_prefixes,
         check_metatile_exists)
 

batch-setup/cron.py

@@ -509,6 +509,8 @@ def is_power_of_two(x):
                         'date prefix, defaults to planet date. You can '
                         'also set the environment variable '
                         'META_DATE_PREFIX.')
+    parser.add_argument('--run-id', help='Distinctive run ID to give to '
+                        'this build. Defaults to planet date YYMMDD.')
 
     args = parser.parse_args()
     planet_date = datetime.strptime(args.date, '%y%m%d')
@@ -525,9 +527,8 @@ def is_power_of_two(x):
     assert args.metatile_size > 0
     assert args.metatile_size < 100
 
-    profile_name = args.profile_name
-    if profile_name is None:
-        profile_name = planet_date.strftime('tps-%y%m%d')
+    run_id = args.run_id or planet_date.strftime('%y%m%d')
+    profile_name = args.profile_name or ('tps-'  + run_id)
 
     def bucket_name(arg_name, bucket_function):
         prop_name = arg_name.lstrip('-').replace('-', '_')
@@ -543,16 +544,15 @@ def bucket_name(arg_name, bucket_function):
     rawr_bucket = bucket_name('--rawr-bucket', 'rawr-tiles')
     missing_bucket = bucket_name('--missing-bucket', 'missing-tiles')
     meta_bucket = bucket_name('--meta-bucket', 'meta-tiles')
-    date_prefix = planet_date.strftime('%y%m%d')
     locations = Locations(
-        Bucket(assets_bucket, 'flat-nodes-' + date_prefix),
-        Bucket(rawr_bucket, date_prefix),
-        Bucket(meta_bucket, date_prefix),
-        Bucket(missing_bucket, date_prefix),
+        Bucket(assets_bucket, 'flat-nodes-' + run_id),
+        Bucket(rawr_bucket, run_id),
+        Bucket(meta_bucket, run_id),
+        Bucket(missing_bucket, run_id),
     )
     meta_date_prefix = (args.meta_date_prefix or
                         os.environ.get('META_DATE_PREFIX') or
-                        date_prefix)
+                        run_id)
 
     iam = boto3.client('iam')
 
@@ -569,9 +569,8 @@ def bucket_name(arg_name, bucket_function):
 
     smgr = boto3.client('secretsmanager')
     smgr_name = (args.db_password_secret_name or
-                 planet_date.strftime('TilesDatabasePassword%y%m%d'))
-    smgr_description = planet_date.strftime(
-        'Tiles database password for %Y-%m-%d import')
+                 ('TilesDatabasePassword' + run_id))
+    smgr_description = 'Tiles database password for %s import' % (run_id,)
     db_password = generate_or_update_password(
         smgr, args.db_password, smgr_name, smgr_description)
 
@@ -585,7 +584,7 @@ def bucket_name(arg_name, bucket_function):
         meta_bucket=locations.meta.name,
         missing_bucket=locations.missing.name,
         date_iso=planet_date.strftime('%Y-%m-%d'),
-        planet_date=planet_date.strftime('%y%m%d'),
+        run_id=run_id,
         raw_tiles_version=args.raw_tiles_version,
         tilequeue_version=args.tilequeue_version,
         vector_datasource_version=args.vector_datasource_version,
@@ -618,7 +617,7 @@ def bucket_name(arg_name, bucket_function):
             ResourceType='instance',
             Tags=[dict(
                 Key='tps-instance',
-                Value=planet_date.strftime('%Y-%m-%d'),
+                Value=run_id,
             )],
         )],
     )

batch-setup/ecr.py

@@ -27,9 +27,8 @@ def ensure_repo(ecr, repo_name):
     return repo_uri
 
 
-def ensure_ecr(planet_date):
+def ensure_ecr(run_id):
     ecr = boto3.client('ecr')
-    date_suffix = planet_date.strftime('%y%m%d')
 
     repo_names = (
         'meta-low-zoom-batch',
@@ -40,7 +39,7 @@ def ensure_ecr(planet_date):
 
     repo_uris = {}
     for repo_name in repo_names:
-        full_name = 'tilezen/%s-%s' % (repo_name, date_suffix)
+        full_name = 'tilezen/%s-%s' % (repo_name, run_id)
         repo_uris[repo_name] = ensure_repo(ecr, full_name)
 
     return repo_uris

batch-setup/make_meta_tiles.py

@@ -1,9 +1,9 @@
 from batch import Buckets
 from batch import run_go
-from datetime import datetime
 import yaml
 from make_rawr_tiles import wait_for_jobs_to_finish
 from make_rawr_tiles import wc_line
+from run_id import assert_run_id_format
 from contextlib import contextmanager
 from collections import namedtuple
 import boto3
@@ -197,9 +197,9 @@ def enqueue_tiles(config_file, tile_list_file, check_metatile_exists):
     parser.add_argument('meta_bucket', help="Bucket with meta tiles in")
     parser.add_argument('--missing-bucket', help="Bucket to store missing "
                         "tile logs in")
-    parser.add_argument('date', help='Planet date, YYMMDD')
+    parser.add_argument('run_id', help='Unique identifier for run.')
     parser.add_argument('--date-prefix', help="Date prefix in bucket, "
-                        "defaults to planet date.")
+                        "defaults to run ID.")
     parser.add_argument('--retries', default=5, type=int, help="Number "
                         "of times to retry enqueueing the remaining jobs "
                         "before giving up.")
@@ -225,11 +225,11 @@ def enqueue_tiles(config_file, tile_list_file, check_metatile_exists):
                         help='Metatile size (in 256px tiles).')
 
     args = parser.parse_args()
-    planet_date = datetime.strptime(args.date, '%y%m%d')
+    assert_run_id_format(args.run_id)
     buckets = Buckets(args.rawr_bucket, args.meta_bucket,
                       args.missing_bucket or args.meta_bucket)
-    date_prefix = args.date_prefix or planet_date.strftime('%y%m%d')
-    missing_bucket_date_prefix = planet_date.strftime('%y%m%d')
+    date_prefix = args.date_prefix or args.run_id
+    missing_bucket_date_prefix = args.run_id
     assert args.key_format_type in ('prefix-hash', 'hash-prefix')
 
     # TODO: split zoom and zoom max should come from config.

batch-setup/make_tiles.py

@@ -4,7 +4,7 @@
 from batch import Buckets
 from batch import create_job_definitions
 from docker import build_and_upload_images
-from datetime import datetime
+from run_id import assert_run_id_format
 import argparse
 import os
 import yaml
@@ -21,7 +21,8 @@ def vpc_of_sg(sg_id):
 
 
 parser = argparse.ArgumentParser('Script to kick off tile creation.')
-parser.add_argument('date', help='Planet date, YYMMDD')
+parser.add_argument('run_id', help='Unique run identifier, used to name '
+                    'resources and log out in the batch runs.')
 parser.add_argument('--num-db-replicas', default=1, type=int,
                     help='Number of database replicas to create.')
 parser.add_argument('rawr_bucket', help='S3 bucket for RAWR tiles')
@@ -30,11 +31,8 @@ def vpc_of_sg(sg_id):
 parser.add_argument('--missing-bucket', default=None,
                     help='Bucket for missing meta tile lists. Defaults to the '
                     'meta bucket.')
-parser.add_argument('--run-id', help='Identifying string to log out in the '
-                    'batch runs.')
 parser.add_argument('--date-prefix', default=None, help='Date prefix to use '
-                    'in S3 buckets. By default, generated from the planet '
-                    'date.')
+                    'in S3 buckets. By default, generated from the run ID.')
 parser.add_argument('--region', help='AWS region. If not provided, then the '
                     'AWS_DEFAULT_REGION environment variable must be set.')
 parser.add_argument('--meta-date-prefix', help='Optional different date '
@@ -44,24 +42,24 @@ def vpc_of_sg(sg_id):
                     'metatile exists first before processing the batch job.')
 
 args = parser.parse_args()
-planet_date = datetime.strptime(args.date, '%y%m%d')
-run_id = args.run_id or planet_date.strftime('%Y%m%d')
-date_prefix = args.date_prefix or planet_date.strftime('%y%m%d')
+run_id = args.run_id
+assert_run_id_format(run_id)
+date_prefix = args.date_prefix or run_id
 
 region = args.region or os.environ.get('AWS_DEFAULT_REGION')
 if region is None:
     import sys
     print "ERROR: Need environment variable AWS_DEFAULT_REGION to be set."
     sys.exit(1)
 
-repo_uris = ensure_ecr(planet_date)
+repo_uris = ensure_ecr(run_id)
 
 # start databases => db_sg & database hostnames
-db_sg_id, database_ids = ensure_dbs(planet_date, args.num_db_replicas)
+db_sg_id, database_ids = ensure_dbs(run_id, args.num_db_replicas)
 
 # create batch environment and job queue
-compute_env_name = planet_date.strftime('compute-env-%y%m%d')
-job_queue_name = planet_date.strftime('job-queue-%y%m%d')
+compute_env_name = 'compute-env-' + run_id
+job_queue_name = 'job-queue-' + run_id
 vpc_id = vpc_of_sg(db_sg_id)
 
 batch_setup(region, vpc_id, [db_sg_id], compute_env_name, job_queue_name)
@@ -92,7 +90,7 @@ def vpc_of_sg(sg_id):
 
 # create job definitions (references databases, batch setup)
 job_def_names = create_job_definitions(
-    planet_date, region, repo_uris, database_ids, buckets, args.db_password,
+    run_id, region, repo_uris, database_ids, buckets, args.db_password,
     memory=memory, vcpus=vcpus, retry_attempts=retry_attempts,
     date_prefix=date_prefix, meta_date_prefix=args.meta_date_prefix,
     check_metatile_exists=args.check_metatile_exists)
@@ -111,7 +109,7 @@ def vpc_of_sg(sg_id):
             'run_id': run_id,
             'retry-attempts': retry_attempts[name],
             'queue-zoom': 7,
-            'job-name-prefix': name + planet_date.strftime('-%y%m%d'),
+            'job-name-prefix': ('%s-%s') % (name, run_id),
             'vcpus': vcpus,
             'job-queue': job_queue_name,
             'job-definition': job_def_names[name],

batch-setup/provision.sh

@@ -42,8 +42,7 @@ export META_BUCKET='%(meta_bucket)s'
 export MISSING_BUCKET='%(missing_bucket)s'
 
 export DATE='%(date_iso)s'
-export PLANET_DATE='%(planet_date)s'
-export DATE_PREFIX='%(planet_date)s'
+export RUN_ID='%(run_id)s'
 export META_DATE_PREFIX='%(meta_date_prefix)s'
 
 export RAW_TILES_VERSION='%(raw_tiles_version)s'
@@ -82,12 +81,12 @@ set -x
 
 python -u /usr/local/src/tileops/import/import.py --find-ip-address meta --date \$DATE \$TILE_ASSET_BUCKET \$AWS_DEFAULT_REGION \
        \$TILE_ASSET_PROFILE_ARN \$DB_PASSWORD
-python -u /usr/local/src/tileops/batch-setup/make_tiles.py --num-db-replicas 10 \$PLANET_DATE --missing-bucket \$MISSING_BUCKET \
+python -u /usr/local/src/tileops/batch-setup/make_tiles.py --num-db-replicas 10 \$RUN_ID --missing-bucket \$MISSING_BUCKET \
        --meta-date-prefix \$META_DATE_PREFIX \$RAWR_BUCKET \$META_BUCKET \$DB_PASSWORD
 python -u /usr/local/src/tileops/batch-setup/make_rawr_tiles.py --config enqueue-rawr-batch.config.yaml --key-format-type hash-prefix \
-       \$RAWR_BUCKET \$DATE_PREFIX \$MISSING_BUCKET
+       \$RAWR_BUCKET \$RUN_ID \$MISSING_BUCKET
 python -u /usr/local/src/tileops/batch-setup/make_meta_tiles.py --date-prefix \$META_DATE_PREFIX --missing-bucket \$MISSING_BUCKET \
-       --key-format-type hash-prefix --metatile-size \$METATILE_SIZE \$RAWR_BUCKET \$META_BUCKET \$DATE_PREFIX
+       --key-format-type hash-prefix --metatile-size \$METATILE_SIZE \$RAWR_BUCKET \$META_BUCKET \$RUN_ID
 EOF
 chmod +x /usr/local/bin/run.sh
 

batch-setup/rds.py

@@ -102,8 +102,8 @@ def set_databases_security_group(rds, security_group_id, databases):
         )
 
 
-def ensure_dbs(planet_date, num_instances):
-    snapshot_id = planet_date.strftime('postgis-prod-%Y%m%d')
+def ensure_dbs(run_id, num_instances):
+    snapshot_id = 'postgis-prod-' + run_id
     rds = boto3.client('rds')
 
     if not does_snapshot_exist(rds, snapshot_id):

batch-setup/run_id.py

@@ -0,0 +1,18 @@
+import re
+import sys
+
+
+def assert_run_id_format(run_id):
+    """
+    Checks that the run ID has a format that means we can use it in resource
+    names, directory and file names, etc... without any problems. Many AWS
+    resources are quite restrictive in terms of the characters that they can
+    contain.
+    """
+
+    m = re.match('^[a-zA-Z0-9]([a-zA-Z0-9-]*[a-zA-Z0-9])?$', run_id)
+    if m is None:
+        print("Run ID %r is badly formed. Run IDs may only contain ASCII "
+              "letters, numbers and dashes. Dashes may not appear at the "
+              "beginning or end of the run ID." % (run_id,))
+        sys.exit(1)