fix: consider to change podman tmp folder location

[alehostert]

Hello there!

I was testing the container snapshot feature (amazing, btw) and got the following error:

write /var/tmp/container_images_storage2299713985/1: no space left on device

Here's the full log:

{
  "id": 15,
  "name": "CreateContainerSnapshotImage",
  "status": "failed",
  "command": "/var/speedia/control container image create-snapshot --container-id 1678b3fdafba --archive-compression-format br",
  "tags": [
    "containerImage"
  ],
  "timeoutSecs": 900,
  "runAt": null,
  "output": "{\"level\":\"error\",\"time\":\"2024-09-24T14:29:52-03:00\",\"error\":{\"error\":\"{\\\"stdErr\\\":\\\"Error: copying layers and metadata for container \\\\\\\"1678b3fdafba77f67ce1c22fe7d8871c767bc644411ad24f67b886d9e712b070\\\\\\\": writing blob: storing blob to file \\\\\\\"/var/tmp/container_images_storage2299713985/1\\\\\\\": write /var/tmp/container_images_storage2299713985/1: no space left on device\\\\n\\\",\\\"exitCode\\\":125}\",\"kind\":\"*infraHelper.CommandError\",\"stack\":null},\"message\":\"CreateContainerSnapshotImageInfraError\"}\n{\"status\":\"infraError\",\"body\":\"CreateContainerSnapshotImageInfraError\"}",
  "err": "{\"stdErr\":\"\",\"exitCode\":1}",
  "startedAt": 1727198486,
  "finishedAt": 1727198992,
  "elapsedSecs": 506,
  "createdAt": 1727198389,
  "updatedAt": 1727198992
}

It's a simple disk space problem, I know. To have the option to change the tmp folder used by podman when generating the snapshot was something to consider, maybe?

Context

My server is using 2 disks:

speedia:~ # lsblk -p
NAME        MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS
/dev/sda      8:0    0 46.6G  0 disk 
├─/dev/sda1   8:1    0    2M  0 part 
├─/dev/sda2   8:2    0   20M  0 part /boot/efi
└─/dev/sda3   8:3    0 46.6G  0 part /usr/local
                                     /srv
                                     /home
                                     /opt
                                     /boot/writable
                                     /boot/grub2/i386-pc
                                     /boot/grub2/x86_64-efi
                                     /.snapshots
                                     /var
                                     /root
                                     /
/dev/sdb      8:16   0  160G  0 disk /var/data

As you can see, the /var/data (sdb) has more space, so, it'll be nice if control was able to detect this and use the bigger partition to host the tmp files during the snapshot generation.

Workaround

There's an easy workaround, tho:

mv /var/tmp/ /var/data/
ln -s /var/data/tmp/ /var/tmp

speedia:~ # ll /var/tmp
lrwxrwxrwx 1 root root 14 Sep 30 14:33 /var/tmp -> /var/data/tmp/

Maybe the KISS principle could be used at this moment and this simple workaround will be enough for now.