storcon: use https for pageservers (#10759)

## Problem

Storage controller uses unsecure http for pageserver API.

Closes: https://github.com/neondatabase/cloud/issues/23734
Closes: https://github.com/neondatabase/cloud/issues/24091

## Summary of changes

- Add an optional `listen_https_port` field to storage controller's Node
state and its API (RegisterNode/ListNodes/etc).
- Allow updating `listen_https_port` on node registration to gradually
add https port for all nodes.
- Add `use_https_pageserver_api` CLI option to storage controller to
enable https.
- Pageserver doesn't support https for now and always reports
`https_port=None`. This will be addressed in follow-up PR.
This commit is contained in:
Dmitrii Kovalkov
2025-02-20 21:16:04 +04:00
committed by GitHub
parent 7c7180a79d
commit e808e9432a
13 changed files with 231 additions and 31 deletions

View File

@@ -1630,6 +1630,7 @@ def neon_env_builder(
class PageserverPort:
pg: int
http: int
https: int | None = None
class LogUtils:
@@ -1886,6 +1887,7 @@ class NeonStorageController(MetricsGetter, LogUtils):
"node_id": int(node.id),
"listen_http_addr": "localhost",
"listen_http_port": node.service_port.http,
"listen_https_port": node.service_port.https,
"listen_pg_addr": "localhost",
"listen_pg_port": node.service_port.pg,
"availability_zone_id": node.az_id,

View File

@@ -3764,3 +3764,56 @@ def test_storage_controller_node_flap_detach_race(
assert len(locs) == 1, f"{shard} has {len(locs)} attached locations"
wait_until(validate_locations, timeout=10)
def test_update_node_on_registration(neon_env_builder: NeonEnvBuilder):
"""
Check that storage controller handles node_register requests with updated fields correctly.
1. Run storage controller and register 1 pageserver without https port.
2. Register the same pageserver with https port. Check that port has been updated.
3. Restart the storage controller. Check that https port is persistent.
4. Register the same pageserver without https port again (rollback). Check that port has been removed.
"""
neon_env_builder.num_pageservers = 1
env = neon_env_builder.init_configs()
env.storage_controller.start()
env.storage_controller.wait_until_ready()
pageserver = env.pageservers[0]
# Step 1. Register pageserver without https port.
env.storage_controller.node_register(pageserver)
env.storage_controller.consistency_check()
nodes = env.storage_controller.node_list()
assert len(nodes) == 1
assert nodes[0]["listen_https_port"] is None
# Step 2. Register pageserver with https port.
pageserver.service_port.https = 1234
env.storage_controller.node_register(pageserver)
env.storage_controller.consistency_check()
nodes = env.storage_controller.node_list()
assert len(nodes) == 1
assert nodes[0]["listen_https_port"] == 1234
# Step 3. Restart storage controller.
env.storage_controller.stop()
env.storage_controller.start()
env.storage_controller.wait_until_ready()
env.storage_controller.consistency_check()
nodes = env.storage_controller.node_list()
assert len(nodes) == 1
assert nodes[0]["listen_https_port"] == 1234
# Step 4. Register pageserver with no https port again.
pageserver.service_port.https = None
env.storage_controller.node_register(pageserver)
env.storage_controller.consistency_check()
nodes = env.storage_controller.node_list()
assert len(nodes) == 1
assert nodes[0]["listen_https_port"] is None