Refactoring DLTJ, Winter 2021 Part 2.5: Fixing the Webmentions Cache

Refactoring DLTJ, Winter 2021 Part 2.5: Fixing the Webmentions Cache

Okay, a half-step backward to fix something I broke yesterday.
As I
described earlier this year
, this static website blog uses the
Webmention protocol
to notify others when I link to their content and receive notifications from others.
Behind the scenes, I’m using the Jekyll plugin called
jekyll-webmention_io
to integrate Webmention data into my blog’s content.
Each time the contents of this site is built, that plug-in contacts the
Webmention.IO service
to receive its Webmention data.
(Webmention.IO holds onto it between Jekyll builds since there is no always-on «dltj.org» server to receive notifications from others.)
The plug-in caches that information to ease the burden on the Webmention.IO service.
The previous CloudFormation-based process was using AWS CodeBuild natively, and the Webmention cache was stored in
CodeBuild’s caching function
.
CodeBuild automatically downloads the previous cache into the working directory for each build iteration and then automatically uploads the cache as the build is completed.
Handy, right?
Well, AWS Amplify simplifies some of the setup of working with the underlying CodeBuild tool.
One of the configuration options that is no longer available is the ability to specify which S3 bucket to use as the CodeBuild cache; so I couldn’t point it at the previous cache files and all of the previous Webmention entries no longer appeared on the blog pages.
Fortunately, I hadn’t decommissioned the CloudFormation stuff, so I still had access to the old cache; I was able to extract the four webmention files (but see below for a discussion about that).
Since Amplify doesn’t allow me to have direct access to the CodeBuild cache, I decided it was high time to use a dedicated cache location for these webmention files.
To do that took three steps:
1. Create the S3 bucket (with no public access)
2. Add read/write policy for that bucket to the AWS role assigned to the Amplify app
3. Add lines to the
amplify.yml
file to copy files from the S3 bucket into and out of the working directory
For step 2, the IAM policy for the Amplify role:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
«Version»
:
«2012-10-17»
,
«Statement»
:
[
{
«Sid»
:
«VisualEditor0»
,
«Effect»
:
«Allow»
,
«Action»
:
[
«s3:DeleteObject»
,
«s3:PutObject»
,
«s3:GetObject»
,
«s3:ListBucket»
],
«Resource»
:
«arn:aws:s3:::org.dltj.webmentions-cache»
},
{
«Sid»
:
«VisualEditor1»
,
«Effect»
:
«Allow»
,
«Action»
:
[
«s3:ListAllMyBuckets»
],
«Resource»
:
«*»
}
]
}
For the
amplify.yml
file:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
version
:
1
frontend
:
phases
:
preBuild
:
commands
:
–
aws s3 cp s3://org.dltj.webmentions-cache webmentions-cache –recursive
–
rvm use $VERSION_RUBY_2_6
–
bundle install –path vendor/bundle
build
:
commands
:
–
rvm use $VERSION_RUBY_2_6
–
bundle exec jekyll build –trace
postBuild
:
commands
:
–
aws s3 cp webmentions-cache …