Skip to content

sighup.md

Latest commit

 

History

History
118 lines (86 loc) · 3.17 KB

sighup.md

File metadata and controls

118 lines (86 loc) · 3.17 KB

SIGHUP behaviour

Logger

Logger level can be reloaded with a SIGHUP.

Prometheus

If any of the fields below are changed, the Prometheus HTTP server is restarted.

prometheus:
  enabled:
  address:
  shutdown_timeout:

pprof

If any of the fields below are changed, the pprof HTTP server is restarted.

pprof:
  enabled:
  address:
  shutdown_timeout:
  enable_block:
  enable_mutex:

enable_block and enable_mutex are applied immediately without restarting the HTTP server.

Object pool

Worker pool sizes are tuned without restarting the pools.

object:
  put:
    pool_size_remote:
  search:
    pool_size:

replicator:
  pool_size:

Policer

Available for reconfiguration fields:

policer:
  head_timeout:
  replication_cooldown:
  object_batch_size:
  boost_multiplier:

Storage engine

Shards can be added, removed or reloaded with SIGHUP. Each shard from the configuration is matched with existing shards by comparing paths from shard.blobstor section. After this we have 3 sets:

  1. Shards that are missing from the configuration (or have mode: disabled) but are currently open. These are closed.
  2. Shards that are added. These are opened and initialized.
  3. Shards that remain in the configuration. For these shards we apply reload to a metabase and for a mode.

Metabase

Changed section Actions
path If path is different, metabase is closed and opened with a new path. All other configuration will also be updated.

FS chain

Changed section Actions
endpoints Updates N3 endpoints.
If new endpoints do not contain the endpoint client is connected to, it will reconnect to another endpoint from the new list. Node service can be interrupted in this case.

Node

Changed section Actions
attribute_* Updates node attributes.

Meta service

fschain:
  endpoints:

The meta service re-applies the updated FS chain RPC endpoints.

gRPC

If any field in the grpc section has changed, all public API gRPC servers are gracefully stopped and restarted with the new configuration. All service implementations (Object, Container, Session, Accounting, Reputation) are re-registered automatically in the new servers.

grpc:
  - endpoint:
    conn_limit:
    tls:
      enabled:
      certificate:
      key:

During the restart there is a short period of unavailability. The control service gRPC endpoint is not affected by SIGHUP.