root/PaintPlus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

README: restore iptables as primary DNS fix with isolation clarification

Browse Source 
iptables fix is simpler (one command, no 13 GB download) and does not
affect container isolation — adds clarifying note so users understand
it only restores Docker's default outbound DNS behaviour.
Docker helper container kept as fallback option.

https://claude.ai/code/session_01WVDg7amsy1TTtxvpku7bcM
This commit is contained in:
Claude
2026-06-14 00:55:09 +00:00
parent d83f94cc24
commit ce29cc1d30
1 changed files with 14 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+14 -10
View File
@@ -210,7 +210,19 @@ If Docker created `./data/` as root and you can't write there without `sudo`, yo
**AI models not downloading (container DNS blocked)** **AI models not downloading (container DNS blocked)**
If the container can't reach HuggingFace (`Errno -3` in logs), use a lightweight Docker helper container to download the models on your behalf — no host Python packages required. The files land in `./data/hf_cache/` which is bind-mounted into the GPU container. If the container can't reach HuggingFace (`Errno -3` in logs), your host firewall is blocking outbound DNS queries from the Docker bridge. The fix below restores Docker's default behaviour — it does **not** affect container isolation (filesystem, network namespace, PID namespace all remain separate):
```bash
sudo iptables -I DOCKER-USER -p udp --dport 53 -j ACCEPT
docker compose -f docker-compose.gpu.yml restart
```
The container will now resolve hostnames and download the models automatically (~13 GB on first run, then cached). Watch progress:
```bash
docker compose -f docker-compose.gpu.yml logs -f | grep -E "local_gpu|Cached|failed"
```
**If you can't run the iptables command**, download with a Docker helper container instead (no host Python needed):
```bash ```bash
# Inpainting model (~6.5 GB) — needed for AI Edit, Make less symmetrical, etc. # Inpainting model (~6.5 GB) — needed for AI Edit, Make less symmetrical, etc.
@@ -230,15 +242,7 @@ docker run --rm \
--exclude '*.msgpack' 'flax_*' 'tf_*'" --exclude '*.msgpack' 'flax_*' 'tf_*'"
``` ```
Once both downloads finish, restart the container: Then restart: `docker compose -f docker-compose.gpu.yml restart`
```bash
docker compose -f docker-compose.gpu.yml restart
```
The first AI Edit request loads from local disk (10–30 s, not a download). Verify in logs:
```bash
docker compose -f docker-compose.gpu.yml logs -f | grep -E "local_gpu|Cached|failed"
```
**Out of VRAM during generation** **Out of VRAM during generation**
- Reduce `LOCAL_GPU_MAX_PIPELINES=1` in `.env` (default 2) - Reduce `LOCAL_GPU_MAX_PIPELINES=1` in `.env` (default 2)