Disk quota exceeded fix pip


Pip

Error

git clone https://github.com/rouralberto/zim-llm.git
./setup.sh

pip install -r requirements.txt

Filesystem      Size  Used Avail Use% Mounted on
/dev/sda1        99G   77G   17G  82% /
devtmpfs        1.9G     0  1.9G   0% /dev
tmpfs           1.9G     0  1.9G   0% /dev/shm
efivarfs         40K   30K  5.4K  85% /sys/firmware/efi/efivars
tmpfs           758M  1.2M  757M   1% /run
tmpfs           1.0M     0  1.0M   0% /run/credentials/systemd-journald.service
tmpfs           1.9G  1.3G  640M  67% /tmp
/dev/sda2      1017M  234M  784M  23% /boot

Downloading torch-2.9.1-cp313-cp313-manylinux_2_28_x86_64.whl (899.7 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╸━━━━━━ 761.0/899.7 MB 11.7 MB/s eta 0:00:12ERROR: Could not install packages due to an OSError: [Errno 122] Disk quota exceeded

   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╸━━━━━━ 761.8/899.7 MB 11.7 MB/s eta 0:00:12
[I] (zim_rag_env) esotericwarfare@arch ~/zim-llm (master) [0|1]>

How can I fix this?

You’re running out of space on /tmp or your user quota, not on the root filesystem as a whole. The key clue is:

ERROR: [Errno 122] Disk quota exceeded

and your /tmp is 67% full (1.3G used) and pip downloads large wheels into /tmp by default.

✔ Why the install is failing

PyTorch wheels are huge (~900 MB). Pip stores the temporary wheel during installation in:

/tmp/pip-*

Your temporary storage or user quota fills up before pip can finish writing the ~900 MB wheel + extracted files.

✅ Fix Options

Option 1 — Tell pip to use a different temp directory

Choose a directory that has plenty of space, e.g. your home folder or /var/tmp.

mkdir -p ~/pip-tmp
TMPDIR=~/pip-tmp pip install -r requirements.txt

or

TMPDIR=/var/tmp pip install -r requirements.txt

Option 2 — Clean /tmp

Check what’s using space:

du -sh /tmp/*

Then remove old temporary files:

sudo rm -rf /tmp/*

(Be careful: only do this if no applications depend on /tmp.)

Option 3 — Clear pip cache

Pip cache can easily grow to several GB.

pip cache purge

Option 4 — Use a smaller PyTorch

If you don’t need CUDA:

pip install torch==2.9.1+cpu --index-url https://download.pytorch.org/whl/cpu

This is much smaller (~150–200 MB instead of ~900 MB).

Option 5 — Increase storage

If you’re on a VPS / container, you may need to expand:

  • The filesystem space
  • Your user quota

👉 Recommended quick fix:

Try this first:

mkdir ~/tmp
TMPDIR=~/tmp pip install -r requirements.txt

If you want, paste your requirements.txt and I can help you optimize or reduce the installed package sizes.