The docker image ls command has an option for filtering output but what if we want to sort the images to find the really hefty ones? You know, the ones that might need to go the way of Old Yeller.
Huh? What’s that, docker? You can’t do that.
If you want something done right…
It ain’t pretty but it works.
docker image ls --format "{{.Repository}}:{{.Tag}} {{.Size}}" | \
awk '{if ($2~/GB/) print substr($2, 1, length($2)-2) * 1000 "MB - " $1 ; else print $2 " - " $1 }' | \
sed '/^0/d' | \
sort -n
Side Note
This assumes that you don’t have images > 999GB.
If you do – I pity you.
Old and busted (docker image ls sample output):
python alpine3.10 992aef1d798f 4 months ago 98.7MB
tomcat-fun alpine 2b7c7bec86f6 5 months ago 140MB
tomcat-fun latest 04af324bcf10 6 months ago 463MB
python latest 7c5fd2af3815 11 months ago 927MB
python alpine3.7 a94f1b57a462 11 months ago 81.1MB
tomcatweather latest 2353bdd18c19 13 months ago 315MB
tomcat latest ca9e2fccef98 13 months ago 463MB
tomcat 9-jre11-slim 8528526b1274 14 months ago 305MB
python v1 76031d3d6d90 14 months ago 74.2MB
python 3.6-alpine a78e257617d1 14 months ago 74.2MB
busybox latest 59788edf1f3e 14 months ago 1.15MB
New hotness (sort script sample output):
1MB - busybox:latest
74MB - python:3.6-alpine
74MB - python:v1
81MB - python:alpine3.7
98MB - python:alpine3.10
140MB - tomcat-fun:alpine
305MB - tomcat:9-jre11-slim
315MB - tomcatweather:latest
463MB - tomcat-fun:latest
463MB - tomcat:latest
927MB - python:latest
Explanation
This monster of Frankenstein needs a proper break down:
1. List docker images with filter
List all images but format the output to just show imageName:tag size.
docker image ls --format "{{.Repository}}:{{.Tag}} {{.Size}}"
2. Convert all image sizes to MB
Check if the size (second argument - $2) is listed in gigabytes ($2~/GB/ checks if GB is in $2) and convert it to megabytes by stripping the size unit (substr($2, 1, length($2)-2)) and multiplying it by 1000. Then append MB and the image name ($1).
awk '{
if ($2~/GB/)
print substr($2, 1, length($2)-2) * 1000 "MB - " $1 ;
If it’s not in gigabytes, assume it’s in megabytes and print out the size and image name.
else
print $2 " - " $1
}'
3. Delete any lines with sizes of 0 (optional)
I had some images that must have broken during creation so they’re listed as 0. Feel free to omit this command.
Look for any line starting with 0 and delete it.
sed '/^0/d'
4. Sort from smallest to largest
The -n flag is the important piece as it sorts it numerically rather than lexicographically/alphabetically. If you prefer, you can also pass the -r flag to reverse sort (largest to smallest).
sort -n
# Lexicographically sorted
109MB - nginx:<none>
109MB - nginx:latest
11.4MB - gcr.io/cloud-builders/metadata:latest
113MB - simple-weather:1
117MB - ubuntu:16.04
1260MB - haskell:8.6
132MB - httpd:2.4
# Numerically sorted
11.4MB - gcr.io/cloud-builders/metadata:latest
109MB - nginx:<none>
109MB - nginx:latest
113MB - simple-weather:1
117MB - ubuntu:16.04
132MB - httpd:2.4
1260MB - haskell:8.6
Quick tip; long explanation.