diff --git a/FAQ.md b/FAQ.md index 5c3401d..c4d7a4b 100644 --- a/FAQ.md +++ b/FAQ.md @@ -113,6 +113,44 @@ To upgrade to `large disk` version, * start master with a larger volume size limit * start volume servers, with reasonable maximum number of volumes +**What's the difference between `large_disk` and `large_disk_full`?** + +- **`large_disk`**: Normal volumes are limited to 30GB each. This `large_disk` build enables support for volumes larger than 30GB (the `5BytesOffset` flag). It uses 17 bytes per file entry instead of 16 bytes, allowing volume sizes up to 8TB. + +- **`large_disk_full`**: This build includes everything in `large_disk`, PLUS additional filer store backend integrations: + - **elastic** - Elasticsearch filer store + - **gocdk** - Go Cloud Development Kit support (for cloud storage backends) + - **rclone** - Rclone integration for remote storage + - **sqlite** - SQLite filer store + - **tarantool** - Tarantool filer store + - **tikv** - TiKV filer store + - **ydb** - YDB (Yandex Database) filer store + + This build is only available for `linux/amd64` due to build complexity. + +**Which one should I use?** + +- Use the **standard (no suffix)** build if your total storage needs are under 30GB per volume and you don't need extra backends. +- Use **`_large_disk`** if you need larger volumes (>30GB each) but don't need the extra filer store backends. +- Use **`_full`** if you need extra filer store backends but don't need large volumes. +- Use **`_large_disk_full`** if you need both large volumes AND extra filer store backends. +- Use **`_large_disk_rocksdb`** if you specifically need RocksDB as a filer store with large disk support. + +**Example Docker commands:** + +```bash +# Standard build +docker run chrislusf/seaweedfs:latest server + +# Large disk build +docker run chrislusf/seaweedfs:3.79_large_disk server + +# Large disk with full backend support +docker run chrislusf/seaweedfs:3.79_large_disk_full server +``` + +**Important:** The `large_disk` versions are NOT compatible with standard 30GB versions. You cannot mix them in the same cluster. See the section above for migration instructions. + ### How large should I configure the volumes? If the system has lots of updates or deletions, it is better to keep the volume size small to reduce compaction load.