docs: initial space for operator docs (#105)

Adding an operator guide to the walrus docs. Pretty light on details
for now, but it can be built out as we get closer to testnet + working
with operators.

---------

Co-authored-by: Markus Legner <[email protected]>

docs/SUMMARY.md

@@ -1,4 +1,5 @@
 <!-- markdownlint-disable headings no-empty-links -->
+
 # Summary
 
 [Walrus](./README.md)
@@ -27,10 +28,13 @@
 
 # Usage
 
-- [Developer Guide](./dev-guide/dev-guide.md)
+- [Developer guide](./dev-guide/dev-guide.md)
   - [Components](./dev-guide/components.md)
   - [Operations](./dev-guide/dev-operations.md)
   - [Sui structures](./dev-guide/sui-struct.md)
+- [Operator guide](./operator-guide/operator-guide.md)
+  - [Storage node](./operator-guide/storage-node.md)
+  - [Aggregator](./operator-guide/aggregator.md)
 - [Setup](./usage/setup.md)
 - [Interacting with Walrus](./usage/interacting.md)
   - [Using the client CLI](./usage/client-cli.md)

docs/dev-guide/dev-guide.md

@@ -1,4 +1,4 @@
-# Developer Guide
+# Developer guide
 
 This guide introduces all the concepts needed to build applications that use Walrus as a storage
 or availability layer. The [overview](../design/overview.md) provides more background and explains

docs/operator-guide/aggregator.md

@@ -0,0 +1,21 @@
+# Operating an aggregator
+
+Below is an example of an aggregator node which hosts a HTTP endpoint that can be used
+to fetch data from Walrus over the web.
+
+The aggregator process is run via the `walrus` client binary in [daemon mode](../usage/web-api.md).
+It can be run in many ways, one example being via a systemd service:
+
+```ini
+[Unit]
+Description=Walrus Aggregator
+
+[Service]
+User=walrus
+Environment=RUST_BACKTRACE=1
+Environment=RUST_LOG=info,walrus=debug
+ExecStart=/opt/walrus/bin/walrus --config /opt/walrus/config/client_config.yaml aggregator --bind-address 0.0.0.0:9000
+Restart=always
+
+LimitNOFILE=65536
+```

docs/operator-guide/operator-guide.md

@@ -0,0 +1,10 @@
+# Operator guide
+
+This chapter introduces all the concepts needed for operators of the different components that make
+up the Walrus system. It is currently a work in progress and will be updated as the platform grows.
+
+Specifically, this guide describes the following:
+
+- [Storage node](storage-node.md) contains the most relevant instructions for operators of a Walrus
+  storage node.
+- [Aggregator](aggregator.md) describes the operation of a Walrus aggregator host.

docs/operator-guide/storage-node.md

@@ -0,0 +1,49 @@
+# Operating a storage node
+
+The binary of the storage node is not yet publicly available. It will be made available in September
+to operators for Testnet nodes. Prior to official network launch the code will be open-sourced.
+
+A basic systemd service running the Storage Node could look like this:
+
+```ini
+[Unit]
+Description=Walrus Storage Node
+
+[Service]
+User=walrus
+Environment=RUST_BACKTRACE=1
+Environment=RUST_LOG=info,walrus=debug
+ExecStart=/opt/walrus/bin/walrus-node run --config-path /opt/walrus/config/walrus-node.yaml
+Restart=always
+
+LimitNOFILE=65536
+```
+
+Make sure to adjust any paths and, if desired, the log level.
+
+The `walrus-node` binary stores slivers in RocksDB, which means the data will be stored on disk, to
+a path configured by the `/opt/walrus/config/walrus-node.yaml` file. The full format with all
+mandatory and optional configuration parameters will be made available with the binary.
+
+Here are some important config params from a shortened version of the `walrus-node.yaml` config
+file:
+
+```yaml
+storage_path: /opt/walrus/db
+metrics_address: 127.0.0.1:9184
+rest_api_address: 0.0.0.0:9185
+sui:
+  rpc: https://fullnode.testnet.sui.io:443
+  system_object: 0xWALRUS_CONTRACT
+blob_recovery:
+  max_concurrent_blob_syncs: 10
+  retry_interval_min_secs: 1
+  retry_interval_max_secs: 3600
+  metadata_request_timeout_secs: 5
+  max_concurrent_metadata_requests: 1
+  sliver_request_timeout_secs: 300
+  invalidity_sync_timeout_secs: 300
+```
+
+For monitoring, you can configure Grafana Agent to fetch metrics from `localhost:9184/metrics`
+(or whatever you've configured `metrics_address` to be).