# Docker-based install can't find its data after upgrade

**URL:** <https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489>\
**Category:** Troubleshooting\
**Created:** [October 28, 2026, 10:45am UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489 "2026-10-28T10:45:38Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![psy-q](https://community.silverbullet.md/user_avatar/community.silverbullet.md/psy-q/32/3731_2.png) [@psy-q](https://community.silverbullet.md/u/psy-q)\
**Post date:** [October 28, 2026, 10:45am UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489/1 "2026-10-28T10:45:38Z")

</div>

When upgrading to `:latest` today I got this message:

```auto
Warning: mounting your space at /space is deprecated, mount the data folder at /data instead

```

So I dutifully did, but after that it still can't find its data and thinks it has to reinitialize an empty instance:

```auto
silverbullet-1 | 2026-10-28T10:35:14.558393Z INFO silverbullet::server: Data folder: /data
silverbullet-1 | 2026-10-28T10:35:14.558977Z INFO silverbullet::boot: boot mode: unconfigured (empty folder, no flags, no legacy env)
silverbullet-1 | 2026-10-28T10:35:14.560033Z INFO silverbullet::boot: SilverBullet setup wizard running: http://0.0.0.0:3000/.setup/

```

I looked through the release notes but the only information on upgrading I ever found was "pull the new image and recreate". What am I missing? Were there migration instructions somewhere?

---

<div class="post-metadata">

**Author:** ![zef](https://community.silverbullet.md/user_avatar/community.silverbullet.md/zef/32/7_2.png) [@zef](https://community.silverbullet.md/u/zef)\
**Post date:** [October 28, 2026, 11:54am UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489/2 "2026-10-28T11:54:59Z")

</div>

What does your volume mapping look like now? It seems to be getting an empty folder mapped at `/data`. Did you volume map your old folder into /data?

---

<div class="post-metadata">

**Author:** ![psy-q](https://community.silverbullet.md/user_avatar/community.silverbullet.md/psy-q/32/3731_2.png) [@psy-q](https://community.silverbullet.md/u/psy-q)\
**Post date:** [October 28, 2026, 12:01pm UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489/3 "2026-10-28T12:01:50Z")

</div>

Yep, I mapped the old dir as it was:

```auto
    volumes:
      - /home/silverbullet/space:/data

```

And that dir has `spaces`, `spaces.json`, `users.json`.

---

<div class="post-metadata">

**Author:** ![zef](https://community.silverbullet.md/user_avatar/community.silverbullet.md/zef/32/7_2.png) [@zef](https://community.silverbullet.md/u/zef)\
**Post date:** [October 28, 2026, 12:08pm UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489/4 "2026-10-28T12:08:36Z")

</div>

Well somehow SilverBullet claims `/data` is empty. Can you somehow get a shell into the running container and check what you see in /data?

---

<div class="post-metadata">

**Author:** ![psy-q](https://community.silverbullet.md/user_avatar/community.silverbullet.md/psy-q/32/3731_2.png) [@psy-q](https://community.silverbullet.md/u/psy-q)\
**Post date:** [October 28, 2026, 12:16pm UTC](https://community.silverbullet.md/t/docker-based-install-cant-find-its-data-after-upgrade/4489/5 "2026-10-28T12:16:53Z")

</div>

Ah, perfect, it seems the files had the wrong owner:

```plaintext
84c499ca667e:/# ls -l /data/
total 12
drwx------ 4 1001 1001 4096 Aug 4 09:21 spaces
-rw------- 1 1001 1001 945 Aug 4 09:21 spaces.json
-rw------- 1 1001 1001 157 Aug 4 09:17 users.json
84c499ca667e:/# ps auxw | grep sb
    1 root 0:00 /sbin/tini -- /docker-entrypoint.sh
   24 root 0:00 grep sb

```

I had `PUID` and `PGID` set to 1000 for some reason. I don't recall it being a problem before and usually I map the UIDs correctly, so not sure what went wrong here, I've been using SB for months before this upgrade.

Anyway, thanks for the pointers, it was definitely an infrastructure issue and not a SB one 🙂
