Image efficiency scanning

[Temikus]

I believe it's worth it to make image efficiency scanning (e.g. dive) a step in CI to make sure we don't bloat the images later down the road.

WDYT?

[lopopolo]

This looks cool. It's a benchmark of sorts. these types of tasks I've put in bench.yaml workflows and they do not block CI. I like running these checks and not blocking on them as a way to measure stats over time.

See https://github.com/artichoke/boba/blob/55124bf7d3e82ef274ceb6dbffe9e6a8a53adbc2/.github/workflows/bench.yaml for an example.

How does one fix a failed check here?

[Temikus]

dive scans for inefficiencies by looking at whether the files get overwritten in multiple layers, wasting overall space.

It does give you a list of files to look at, e.g.:

λ dive --ci  haugene/transmission-openvpn
Using config file: /Users/temikus/.dive.yaml
  Using default CI config
Image Source: docker://haugene/transmission-openvpn
Fetching image... (this can take a while for large images)
Analyzing image...
  efficiency: 96.9607 %
  wastedBytes: 18154395 bytes (18 MB)
  userWastedPercent: 6.7145 %
Inefficient Files:
Count  Wasted Space  File Path
    2        3.2 MB  /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.25
    2        1.3 MB  /var/cache/debconf/templates.dat
    2        834 kB  /lib/x86_64-linux-gnu/libfdisk.so.1.1.0
    2        742 kB  /var/lib/dpkg/info/libstdc++6:amd64.symbols
    2        680 kB  /lib/x86_64-linux-gnu/libmount.so.1.1.0
    2        623 kB  /lib/x86_64-linux-gnu/libblkid.so.1.1.0
    2        562 kB  /var/cache/apt/pkgcache.bin
    2        484 kB  /var/log/dpkg.log
    2        425 kB  /var/cache/apt/srcpkgcache.bin
    2        392 kB  /lib/x86_64-linux-gnu/libsmartcols.so.1.1.0
    2        336 kB  /var/lib/dpkg/status
    2        296 kB  /var/log/lastlog
    2        250 kB  /sbin/fdisk
    2        217 kB  /sbin/sfdisk
    2        193 kB  /sbin/cfdisk
    2        193 kB  /lib/x86_64-linux-gnu/libgcc_s.so.1
    2        185 kB  /sbin/zramctl
    2        184 kB  /sbin/fsck.minix
    2        176 kB  /sbin/blkid
    2        176 kB  /usr/bin/partx
    2        168 kB  /sbin/losetup
    2        168 kB  /bin/lsblk
    2        160 kB  /sbin/mkfs.minix
    2        160 kB  /sbin/mkswap
    2        144 kB  /bin/dmesg
    2        144 kB  /usr/bin/lsipc
    2        144 kB  /usr/bin/lscpu
    2        136 kB  /usr/share/gcc-8/python/libstdcxx/v6/printers.py
    2        130 kB  /bin/findmnt
    2        128 kB  /usr/share/doc/gcc-8-base/copyright
    2        127 kB  /usr/bin/lslogins
 ...
    2           0 B  /usr/share/doc/mount/mount.txt
    2           0 B  /sbin/getty
    2           0 B  /usr/share/doc/util-linux/changelog.Debian.gz
    2           0 B  /usr/share/doc/fdisk/changelog.Debian.gz
Results:
  PASS: highestUserWastedPercent
  SKIP: highestWastedBytes: rule disabled
  PASS: lowestEfficiency
Result:PASS [Total:3] [Passed:2] [Failed:0] [Warn:0] [Skipped:1]

We can run it both as a benchmark and a gate (e.g. prevent us from wasting 50% image space or something high like that).

[lopopolo]

Sounds great let's do it.