Logger level can be reloaded with a SIGHUP.
Available for reconfiguration fields:
head_timeout:
replication_cooldown:
object_batch_size:
max_workers:
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:
- Shards that are missing from the configuration (or have
mode: disabled
) but are currently open. These are closed. - Shards that are added. These are opened and initialized.
- Shards that remain in the configuration.
For these shards we apply reload to a
metabase
only. Ifresync_metabase
is true, the metabase is also resynchronized.
Changed section | Actions |
---|---|
path |
If path is different, metabase is closed and opened with a new path. All other configuration will also be updated. |
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. |