Podman always uses /var/tmp for loading docker images and not removing the files after loading

[PieperCH]

Is this a BUG REPORT or FEATURE REQUEST? (leave only one on its own line)

/kind bug

Description

When using podman load with image on commandline it always creates temporary files in /var/tmp. This is ok for temporary usage, but as podman doesn't remove the file afterwards. It's also hardcoded to use /var/tmp for loading the files.
As a result /var/tmp is running out of memory when loading a lot of images
Giving TMPDIR bash variable and --tmpdir doesn't help, as it seems to be hardcoded in load.go in line 78.

Steps to reproduce the issue:

1. Set variable TMPDIR to some other directory <tmpdir>

2. Load a .tar.gz into storage with podman load --tmpdir=<tmpdir> -i <filename>

3. Check /var/tmp directory for files called docker-tar*

Describe the results you received:
/var/tmp is running out of memory when loading a lot of images

Describe the results you expected:
Two options:
1.) Delete /var/tmp/docker-tar file after loading
2.) Give the possibility to specify another directory

Additional information you deem important (e.g. issue happens only occasionally):

Output of podman version:
Version: 1.4.4
RemoteAPI Version: 1
Go Version: go1.10.3
OS/Arch: linux/amd64

Output of podman info --debug:
debug:
compiler: gc
git commit: ""
go version: go1.10.3
podman version: 1.4.4
host:
BuildahVersion: 1.9.0
Conmon:
package: podman-1.4.4-4.el7.x86_64
path: /usr/libexec/podman/conmon
version: 'conmon version 0.3.0, commit: unknown'
Distribution:
distribution: '"rhel"'
version: "7.7"
MemFree: 1229807616
MemTotal: 16647565312
OCIRuntime:
package: runc-1.0.0-65.rc8.el7.x86_64
path: /usr/bin/runc
version: 'runc version spec: 1.0.1-dev'
SwapFree: 5368172544
SwapTotal: 5368705024
arch: amd64
cpus: 4
hostname: d-sccp-pod-1.draco-242.sccloudres.net
kernel: 3.10.0-1062.12.1.el7.x86_64
os: linux
rootless: true
uptime: 20h 21m 44.21s (Approximately 0.83 days)
registries:
blocked: null
insecure: null
search:

• registry.access.redhat.com
• docker.io
• registry.fedoraproject.org
• quay.io
• registry.centos.org
  store:
  ConfigFile: /home/sccp/.config/containers/storage.conf
  ContainerStore:
  number: 0
  GraphDriverName: vfs
  GraphOptions: null
  GraphRoot: /pod/sccp/.local/share/containers/storage
  GraphStatus: {}
  ImageStore:
  number: 2
  RunRoot: /pod/sccp/tmp/run-1002
  VolumePath: /pod/sccp/.local/share/containers/storage/volumes

Package info (e.g. output of rpm -q podman or apt list podman):

podman-1.4.4-4.el7.x86_64

Additional environment details (AWS, VirtualBox, physical, etc.):

[karlyeurl]

It's amusing that I was preparing a similar bug report, because I encountered yesterday the same issue in an environment where /var is on a ridiculously small partition.

I would like to point out that libpod/runtime_img.go also contains two ocurrences of ioutil.TempFile with a hardcoded reference to /var/tmp.

If there is no good reason to force these values, I can make a PR with the necessary changes.

(Note: github.com/containers/image/v5 also refers to this directory for the builtinUnixTempDirForBigFiles, I wonder if that needs to be looked at as well.)

[vincentbernat]

Half of this bug is now fixed (we can choose another directory). The other half is not: after loading an image, the docker-tar* files are kept in the temporary location.

[rhatdan]

Could you give me an exact reproducer?
And show me the left over files.

[rhatdan]

BTW You should open a new issue not add to a closed issue.