你好，我在一台运行 Debian 13 的 25GB 云虚拟机上托管了一个小型自托管论坛。我尝试使用 ./launcher rebuild app 更新到 2026.6.0-latest，但因磁盘空间不足而取消。我按照提示清理了 Docker 镜像，但问题仍未解决。目前剩余空间为 3.9 GB。我的论坛上传文件总计 138 MB，备份文件总计 448 MB。除了 Discourse 安装外，服务器上没有其他内容，我也未进行其他操作。
在使用 du 命令深入排查后，我发现 /var/lib/containerd 占用了 13 GB，而 /var/docker 占用了 5.1 GB。这些目录占用如此多空间是否正常？有什么方法可以处理吗？
如果这是正常现象，那我可能只能付费增加存储了。在承担额外费用之前，我想先确认一下。谢谢！
嗯……你试过 ./launcher cleanup 吗？
运行清理是值得一试的。请发布具体发生的情况。这是我曾经做过的一次清理：
root@ubuntu-2gb-nbg1-1:/var/discourse# df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 19G 12G 6.5G 65% /
root@ubuntu-2gb-nbg1-1:/var/discourse# ./launcher cleanup
WARNING! This will remove all stopped containers.
Are you sure you want to continue? [y/N] y
Total reclaimed space: 0B
WARNING! This will remove all images without at least one container associated to them.
Are you sure you want to continue? [y/N] y
Deleted Images:
deleted: sha256:33ce56b3484175342191a344fe1aa9f330431de7bf6f0f0e14f75c5bd851a5b0
deleted: sha256:5e04a0411f813e8f2c1a50d84a7d7bb1b17a855ce041730cf8e3b2fb90144b19
untagged: discourse/base:2.0.20200512-1735
untagged: discourse/base@sha256:7f6c5be23a8e4237cecafaca9d041de5964f8237345b7b183cebdee1f73ed024
deleted: sha256:991acdba0b1f18141192f8682bc5ec1c3365dff0d4fa1c6504e6ce0e3ae6dd76
deleted: sha256:e1c41c8dbd697ae0a61ca612c61f085f69f5bf74c4269cf5c68b4de0fec723fa
deleted: sha256:c2adabaecedbda0af72b153c6499a0555f3a769d52370469d8f6bd6328af9b13
untagged: discourse/base:2.0.20200724-1815
untagged: discourse/base@sha256:1d5eb2f8eae84203b28d3c643648d0dd7d213c2055865cfa5132e30db9e365d5
Total reclaimed space: 2.591GB
# df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 19G 8.7G 9.3G 49% /
也值得详细查看磁盘使用情况并在此发布！例如参见：
如果系统上有备份，请下载它们、测试它们，并删除除最新备份外的所有备份。如果处于极端压力下，连最新备份也删除。（在我的情况下，我确保下载的副本首先由我个人的异地备份覆盖。始终至少保留两份副本。）
除了 du（它在操作系统层面回答问题），你还可以在 Docker 层面提出问题。请发布你的详细发现！
# docker volume ls -qf dangling=true
# docker images -a
在 25G 的实例上管理磁盘使用量很困难——我这样做过很长时间，耗费了时间并利用了系统管理员的经验。最终我更换了服务提供商，花费更少并获得了 40G，情况好多了。
另请参阅之前的内容：
Discourse 升级因 25G droplet 磁盘空间不足而失败
这非常有信息量，谢谢。
在我尝试 cleanup 之前，Discourse 是否提供了一种无需执行恢复即可验证备份完整性的方法？
du -hx / | sort -h | tail -49 的输出结果（我发现 -h 选项让结果更容易解析）：
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle/ruby
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle/ruby/3.4.0
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www/discourse
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www/discourse/vendor
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www/discourse/vendor/bundle
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www/discourse/vendor/bundle/ruby
837M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs/var/www/discourse/vendor/bundle/ruby/3.4.0
839M /var/discourse/shared
839M /var/discourse/shared/standalone
840M /var/discourse
863M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42
863M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs
868M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283
868M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/283/fs
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/284
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/284/fs
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs
908M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/263/fs/usr
925M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22/fs/usr
944M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/263
944M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/263/fs
961M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22
961M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22/fs
1.3G /usr
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var/www
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var/www/discourse
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www/discourse
1.6G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227
1.6G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs
2.5G /var/log
2.5G /var/log/journal
2.5G /var/log/journal/1d9299c9656046d6bd0e18d4a09c183c
2.7G /var/lib/containerd/io.containerd.content.v1.content
2.7G /var/lib/containerd/io.containerd.content.v1.content/blobs
2.7G /var/lib/containerd/io.containerd.content.v1.content/blobs/sha256
9.8G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs
9.8G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots
13G /var/lib
13G /var/lib/containerd
17G /var
20G /
docker volume ls -qf dangling=true 的输出为空。实际上，docker volume ls 也是空的。
docker images -a 的输出：
IMAGE ID DISK USAGE CONTENT SIZE EXTRA
discourse/base:2.0.20260209-1300 50d2dae489a9 4.77GB 1.12GB
discourse/base:2.0.20260521-0047 27373b823bbe 4.89GB 1.17GB
local_discourse/app:latest 01527763be8d 6.9GB 1.63GB U
我对 Docker 的了解不足以诊断这些输出，但如果 /var/log/ 仅包含日志，其大小似乎有点大。另外，我只有一个论坛在运行，却有三个镜像，这很有趣（当然，这也可能是完全正常的）。
感谢您关于增加存储空间的建议。我可能会进行扩容，但在这样做之前，这也是一次很好的学习经历。
你可以通过重新配置日志记录来节省一些空间：
我其实一直不太理解 Docker 容器的管理。有时候清理会有帮助。我刚才运行了以下命令：
# cd /var/discourse/
# df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 38G 24G 13G 67% /
# docker images -a
i Info → U In Use
IMAGE ID DISK USAGE CONTENT SIZE EXTRA
discourse/base:2.0.20260209-1300 18ccefc73392 3.1GB 0B
local_discourse/app:latest 810ac5579584 4.57GB 0B U
<untagged> d303cfc49c99 4.37GB 0B
<untagged> a9d03385c205 4.34GB 0B
# ./launcher cleanup
WARNING! This will remove all stopped containers.
Are you sure you want to continue? [y/N] y
Total reclaimed space: 0B
WARNING! This will remove all images without at least one container associated to them.
Are you sure you want to continue? [y/N] y
Deleted Images:
deleted: sha256:d303cfc49c992b7fa0b0da65e38f3de79312c6e42eb99e5fd7b27bf4e237ab1d
deleted: sha256:829057c668fd2bd0f7b863c2589e5c3c27364285b39d6df31d38eb208e6410df
untagged: discourse/base:2.0.20260209-1300
untagged: discourse/base@sha256:50d2dae489a9c9916c3d50d880f7f1282fcbe179339b60a8947ab135c1f1f9bf
deleted: sha256:a9d03385c205f740e5f02db0c665a83688f7194d5386bd2d01772b8b5cd5dc65
deleted: sha256:70ec1b4b6cab7365b5f56b8187ef7631e6906fe13b5d3a2b1e39d3457d14514c
Total reclaimed space: 2.517GB
# docker images -a
i Info → U In Use
IMAGE ID DISK USAGE CONTENT SIZE EXTRA
local_discourse/app:latest 810ac5579584 4.57GB 0B U
<untagged> 18ccefc73392 3.1GB 0B
# df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 38G 21G 15G 59% /
./launcher cleanup 的输出：
警告！这将移除所有已停止的容器。
您确定要继续吗？[y/N] y
总共回收空间：0B
警告！这将移除所有未关联任何容器的镜像。
您确定要继续吗？[y/N] y
已删除的镜像：
untagged: discourse/base:2.0.20260209-1300
deleted: sha256:50d2dae489a9c9916c3d50d880f7f1282fcbe179339b60a8947ab135c1f1f9bf
deleted: sha256:1bb6f08ca8363e7e002eaf8bafc76b7742d8baa429f816cde77d3d38d234c98a
deleted: sha256:07596c600885cf3f0647c4b41959c5631328ee37af14913b9f2b1b1771ad4f45
untagged: discourse/base:2.0.20260521-0047
deleted: sha256:27373b823bbeca526dd006cefe1c58e7eeca1e5670ab6e619fcd2138f7d50420
deleted: sha256:ab90467600edc09bae8ac2f43420b7c771f87d357aa15646cd40025ec0774fda
deleted: sha256:053438a5cf1727f724f96a4d7af0b135929f633afd8ee05edcddeaf99232b636
总共回收空间：1.17GB
不过有点奇怪：我的 Discourse 论坛刚刚从显示剩余 3.9 GB 变为剩余 8.5 GB。如果我再次运行 du -hx / | sort -h | tail -49，您可以与我之前的结果进行对比，/var/lib 已经减少了 8 GB：
628M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www/discourse/vendor/bundle/ruby
628M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www/discourse/vendor/bundle/ruby/3.4.0
629M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www/discourse/vendor
733M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle/ruby/3.4.0/gems
792M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs/var/www/discourse/node_modules/.pnpm
793M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs/var/www/discourse/node_modules
794M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs/var
794M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs/var/www
794M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs/var/www/discourse
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle/ruby
833M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs/var/www/discourse/vendor/bundle/ruby/3.4.0
839M /var/discourse/shared
839M /var/discourse/shared/standalone
841M /var/discourse
863M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42
863M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/42/fs
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43
874M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/43/fs
925M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22/fs/usr
961M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22
961M /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/22/fs
1.3G /usr
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var/www
1.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/251/fs/var/www/discourse
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www
1.5G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs/var/www/discourse
1.6G /var/lib/containerd/io.containerd.content.v1.content
1.6G /var/lib/containerd/io.containerd.content.v1.content/blobs
1.6G /var/lib/containerd/io.containerd.content.v1.content/blobs/sha256
1.6G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227
1.6G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/227/fs
2.5G /var/log
2.5G /var/log/journal
2.5G /var/log/journal/1d9299c9656046d6bd0e18d4a09c183c
6.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs
6.4G /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots
7.9G /var/lib/containerd
8.0G /var/lib
12G /var
15G /
此外，docker images -a 的输出也有明显不同：
镜像 ID 磁盘使用量 内容大小 额外信息
local_discourse/app:latest 01527763be8d 6.9GB 1.63GB U
看起来 ./launcher cleanup 报告的回收空间偏少？
我现在应该有足够空间执行更新了。看来在更新后运行 cleanup 是一个好习惯。
感谢您的帮助！
您刚刚清理了两个基础镜像，保留了本地构建的镜像。
基础镜像只是原始安装过程的残留物吗？还是说在执行 rebuild app 时也会下载它们？
是的。如果缺少最新镜像，将会自动下载，并以此为基础构建您的本地镜像。
我想不出有什么理由需要保留多个基础镜像……