add support for cdn nodes

OriHoch · OriHoch · commit a612a7ba1359 · 2024-02-25T18:51:48.000+02:00
diff --git a/cwm_worker_operator/cli.py b/cwm_worker_operator/cli.py
@@ -1,5 +1,7 @@
 import sys
 import importlib
+import subprocess
+
 import click
 
 
@@ -187,5 +189,15 @@ def start_minio_auth_server_devel():
     uvicorn.run('cwm_worker_operator.minio_auth_plugin.app:app', host='0.0.0.0', port=5000, reload=True)
 
 
+@main.command(short_help="Start consecutive daemon run once commands")
+@click.argument('DAEMON_NAME', nargs=-1)
+def multi_run_once(daemon_name):
+    for name in daemon_name:
+        print(f'Running {name} --run-once')
+        subprocess.check_call([
+            'cwm-worker-operator', name, 'start_daemon', '--once'
+        ])
+
+
 if __name__ == '__main__':
     main()
diff --git a/cwm_worker_operator/deployments_manager.py b/cwm_worker_operator/deployments_manager.py
@@ -267,9 +267,9 @@ def delete(self, namespace_name, deployment_type, delete_data=False, **kwargs):
             minio_admin = get_minio_admin()
             failed = False
             for cmd, args, kwargs in [
-                ('user_remove', [namespace_name], {}),
                 ('policy_unset', [namespace_name], {'user': namespace_name}),
                 ('policy_remove', [namespace_name], {}),
+                ('user_remove', [namespace_name], {}),
             ]:
                 try:
                     getattr(minio_admin, cmd)(*args, **kwargs)
@@ -328,16 +328,21 @@ def iterate_cluster_nodes(self):
                 assert ret == 0, out
                 node = json.loads(out)
                 is_worker = False
+                is_cdn = False
                 for taint in node.get('spec', {}).get('taints', []):
-                    if taint.get('key') == 'cwmc-role'and taint.get('value') == 'worker':
-                        is_worker = True
+                    if taint.get('key') == 'cwmc-role':
+                        if taint.get('value') == 'worker':
+                            is_worker = True
+                        elif taint.get('value') == 'cdn':
+                            is_cdn = True
                         break
                 unschedulable = bool(node.get('spec', {}).get('unschedulable'))
                 public_ip = node.get('status', {}).get('addresses', [{}])[0].get('address', '')
                 labels = node.get('metadata', {}).get('labels', {})
                 yield {
                     'name': node_name,
                     'is_worker': is_worker,
+                    'is_cdn': is_cdn,
                     'unschedulable': unschedulable,
                     'public_ip': public_ip,
                     'cleaner_cordoned': labels.get(NODE_CLEANER_CORDON_LABEL) == 'yes'
@@ -376,12 +381,19 @@ def iterate_dns_healthchecks(self):
                         healthcheck_name = tag['Value']
                         break
                 healthcheck_ip = healthcheck.get('HealthCheckConfig', {}).get('IPAddress') or ''
-                if healthcheck_name and healthcheck_name.startswith(config.DNS_RECORDS_PREFIX + ":"):
-                    yield {
-                        'id': healthcheck_id,
-                        'node_name': healthcheck_name.replace(config.DNS_RECORDS_PREFIX + ":", ""),
-                        'ip': healthcheck_ip
-                    }
+                if healthcheck_name:
+                    if healthcheck_name.startswith(config.DNS_RECORDS_PREFIX + ":"):
+                        yield {
+                            'id': healthcheck_id,
+                            'node_name': healthcheck_name.replace(config.DNS_RECORDS_PREFIX + ":", ""),
+                            'ip': healthcheck_ip
+                        }
+                    elif healthcheck_name.startswith('minio-' + config.DNS_RECORDS_PREFIX + ":"):
+                        yield {
+                            'id': healthcheck_id,
+                            'node_name': healthcheck_name.replace('minio-' + config.DNS_RECORDS_PREFIX + ":", ""),
+                            'ip': healthcheck_ip
+                        }
             if res['IsTruncated']:
                 next_marker = res['NextMarker']
             else:
@@ -398,7 +410,10 @@ def iterate_dns_records(self):
                 **({'StartRecordType': next_record_type} if next_record_type is not None else {}),
             )
             for record in res.get('ResourceRecordSets', []):
-                if record['Type'] == 'A' and record['Name'] == '{}.{}.'.format(config.DNS_RECORDS_PREFIX, config.AWS_ROUTE53_HOSTEDZONE_DOMAIN):
+                if record['Type'] == 'A' and (
+                    record['Name'] == '{}.{}.'.format(config.DNS_RECORDS_PREFIX, config.AWS_ROUTE53_HOSTEDZONE_DOMAIN)
+                    or record['Name'] == 'minio-{}.{}.'.format(config.DNS_RECORDS_PREFIX, config.AWS_ROUTE53_HOSTEDZONE_DOMAIN)
+                ):
                     if record['SetIdentifier'].startswith(config.DNS_RECORDS_PREFIX+':'):
                         record_ip = ''
                         for value in record.get('ResourceRecords', []):
@@ -423,7 +438,7 @@ def set_dns_healthcheck(self, node_name, node_ip):
             CallerReference=caller_reference,
             HealthCheckConfig={
                 "IPAddress": node_ip,
-                "Port": 12808,
+                "Port": 80,
                 "Type": "HTTP",
                 "ResourcePath": "/healthz",
                 "RequestInterval": 30,  # according to AWS docs, when using the recommended regions, it actually does a healthcheck every 2-3 seconds
@@ -440,17 +455,20 @@ def set_dns_healthcheck(self, node_name, node_ip):
         )
         return healthcheck_id
 
-    def set_dns_record(self, node_name, node_ip, healthcheck_id):
+    def set_dns_record(self, node_name, node_ip, healthcheck_id, node_type):
         client = boto3.client('route53')
+        prefix = ''
+        if node_type == 'worker':
+            prefix = 'minio-'
         client.change_resource_record_sets(
             HostedZoneId=config.AWS_ROUTE53_HOSTEDZONE_ID,
             ChangeBatch={
-                "Comment": "cwm-worker-operator deployments_manager.set_dns_record({},{})".format(node_name, node_ip),
+                "Comment": "cwm-worker-operator deployments_manager.set_dns_record({}{},{})".format(prefix, node_name, node_ip),
                 "Changes": [
                     {
                         "Action": "CREATE",
                         "ResourceRecordSet": {
-                            "Name": '{}.{}.'.format(config.DNS_RECORDS_PREFIX, config.AWS_ROUTE53_HOSTEDZONE_DOMAIN),
+                            "Name": '{}{}.{}.'.format(prefix, config.DNS_RECORDS_PREFIX, config.AWS_ROUTE53_HOSTEDZONE_DOMAIN),
                             'Type': 'A',
                             'SetIdentifier': config.DNS_RECORDS_PREFIX + ':' + node_name,
                             'MultiValueAnswer': True,
diff --git a/cwm_worker_operator/nodes_checker.py b/cwm_worker_operator/nodes_checker.py
@@ -20,20 +20,25 @@
 
 def set_node_healthy_keys(domains_config, deployments_manager):
     healthy_node_name_ips = {}
+    node_name_types = {}
     # set healthy redis key of healthy nodes - so that ingress will report them as healthy
     for node in deployments_manager.iterate_cluster_nodes():
         if node['is_worker']:
+            node_name_types[node['name']] = 'worker'
+        elif node['is_cdn']:
+            node_name_types[node['name']] = 'cdn'
+        if node['name'] in node_name_types:
             healthy_node_name_ips[node['name']] = node['public_ip']
             domains_config.set_node_healthy(node['name'], True)
     # del healthy redis key for nodes which have a key but are not in list of healthy nodes
     for node_name in domains_config.iterate_healthy_node_names():
         if node_name not in healthy_node_name_ips:
             domains_config.set_node_healthy(node_name, False)
-    return healthy_node_name_ips
+    return healthy_node_name_ips, node_name_types
 
 
-def update_dns_records(deployments_manager, healthy_node_name_ips):
-    logs.debug('update_dns_records', healthy_node_name_ips=healthy_node_name_ips, debug_verbosity=10)
+def update_dns_records(deployments_manager, healthy_node_name_ips, node_name_types):
+    logs.debug('update_dns_records', healthy_node_name_ips=healthy_node_name_ips, node_name_types=node_name_types, debug_verbosity=10)
     dns_healthchecks = {}
     dns_records = {}
     # collect node names of existing dns healthchecks / records
@@ -43,8 +48,7 @@ def update_dns_records(deployments_manager, healthy_node_name_ips):
     for dns_healthcheck in deployments_manager.iterate_dns_healthchecks():
         logs.debug('dns_healthcheck', dns_healthcheck=dns_healthcheck, debug_verbosity=10)
         dns_healthchecks[dns_healthcheck['node_name']] = {'id': dns_healthcheck['id'], 'ip': dns_healthcheck['ip']}
-        if dns_healthcheck['node_name'] in healthy_node_name_ips and dns_healthcheck['ip'] != healthy_node_name_ips[
-            dns_healthcheck['node_name']]:
+        if dns_healthcheck['node_name'] in healthy_node_name_ips and dns_healthcheck['ip'] != healthy_node_name_ips[dns_healthcheck['node_name']]:
             deployments_manager.delete_dns_healthcheck(dns_healthcheck['id'])
             update_required_healthcheck_node_names.add(dns_healthcheck['node_name'])
     for dns_record in deployments_manager.iterate_dns_records():
@@ -65,7 +69,7 @@ def update_dns_records(deployments_manager, healthy_node_name_ips):
         else:
             healthcheck_id = dns_healthchecks[node_name]['id']
         if node_name not in dns_records or node_name in update_required_records_node_names:
-            deployments_manager.set_dns_record(node_name, node_ip, healthcheck_id)
+            deployments_manager.set_dns_record(node_name, node_ip, healthcheck_id, node_name_types[node_name])
     return dns_healthchecks, dns_records
 
 
@@ -81,8 +85,8 @@ def delete_dns_records(deployments_manager, healthy_node_name_ips, dns_healthche
 
 
 def run_single_iteration(domains_config, deployments_manager, **_):
-    healthy_node_name_ips = set_node_healthy_keys(domains_config, deployments_manager)
-    dns_healthchecks, dns_records = update_dns_records(deployments_manager, healthy_node_name_ips)
+    healthy_node_name_ips, node_name_types = set_node_healthy_keys(domains_config, deployments_manager)
+    dns_healthchecks, dns_records = update_dns_records(deployments_manager, healthy_node_name_ips, node_name_types)
     delete_dns_records(deployments_manager, healthy_node_name_ips, dns_healthchecks, dns_records)
 
 
diff --git a/cwm_worker_operator/updater.py b/cwm_worker_operator/updater.py
@@ -138,6 +138,7 @@ def get_instances_updates(domains_config: DomainsConfig, cwm_api_manager: CwmApi
     last_update = None
     instances_updates = {}
     for update in cwm_api_manager.get_cwm_updates(from_datetime):
+        logs.debug(f'get_instances_updates: update={update}', debug_verbosity=10)
         if last_update is None or last_update < update['update_time']:
             last_update = update['update_time']
         if update['worker_id'] not in instances_updates:
