π
π
This is partially because users are being migrated to @us-chi1.cove.seamark.dev and that currently doesnβt have bluesky notifications turned on.
Behind the scenes, the databases and embedded pds now support libsql (aka turso).
its now fully using the AWS SDK v2 instead of the v1 shipped with distribution library.
Online Bluesky US-East https://relay1.us-east.bsky.network requestCrawl listReposByCollection Known (rev: 3mezn66m4422...) Online Bluesky US-West https://relay1.us-west.bsky.network requestCrawl listReposByCollection Known (rev: 3mezn66m4422...) Online Firehose NA https://northamerica.firehose.network requestCrawl Known (rev: 3mezn66m4422...) Online Firehose EU https://europe.firehose.network requestCrawl Known (rev: 3mezn66m4422...) Online Firehose Asia https://asia.firehose.network requestCrawl Known (rev: 3mezn66m4422...) Online Microcosm Montreal https://relay.fire.hose.cam requestCrawl Known (rev: 3mezn66m4422...)
Other changes are the Hold now supports did:plc instead of just did:web. This will allow easier migrations across domains.
The admin panel also has a new relay page to check if the relays have the latest data. this is inspired by debug.hose.cam but contains the firehose.network relays as well.
Vulnerability Scan Results 1298 vulnerabilities found 16 138 304 39 List of various CVEs and severity levels for packages.
Big changes happening.
Vulnerability scanning has (mostly) landed. you can take a look at the example here atcr.io/r/evan.jarre...
It will be processing images in the background and eventually on push. This is built using Syft and Grype. and records are stored in the holds embedded pds
Haven't posted in awhile but that doesn't mean big things aren't being done.
Lots of changes coming soon!
There are 3 modes of operation.
In order to push you have to have an atproto account.
Then for pulls you can decide between:
specific accounts
all accounts
anonymous (no account required)
There is a setting for running your own, which denies anyone but you push/pulling but the manifests would still be public, so people could see the name/tag, but not download the image. Image layers are still off protocol since they can get too big for the pds.
Tfw you already follow a container registry on atproto
Tangled.org commit history: jk found more fixes 865c5971 evan.jarrett.net 16 minutes ago i don't think i can make this website any faster... 536fa416 evan.jarrett.net 50 minutes ago
Spent the weekend micro-optimizing everything on the atcr.io
page rendering and navigation should be much faster, and avatars are now not trying to load the full native resolution.
showing the new hero banner for atcr.io. your registry at sea docker login atcr.io docker push atcr.io/your/repo cute little manatee mascot off to the right
Spent some time on the UI. The homepage got more of a redesign.
Also added the ability for data export and deletions, so data that is stored by ATCR, but not in your PDS can be viewed/removed at any time.
As a funny note, the real storage is closer to 11GB so far.
A bug was found in the process where some pushes were not properly recording the layer size. this means those accounts get free storage until a new push re-calculates the usage
:)
Hold Admin Dashboard | Crew | Settings Dashboard Crew Members: 55 Storage: 7.2 GB 378 unique blobs Tier Distribution: deckhand 55 members Top Users by Storage Members | Usage | Blobs
In addition to working on quotas, An admin dashboard was added to Holds to manage access, quotas, view metrics etc.
This is all running off public data from PDS, but gives an easy way to view/edit.
As a reminder, you own your manifests, tags, descriptions, avatars and more in your PDS. But due to limits on blob sizes, layers/blobs are stored in S3, and retrieved via a sidecar service. Anyone can run their own Hold with a docker container and S3 compatible storage.
Screenshot from atcr.io showing quotas from settings page: Stowage Estimated storage usage on your default hold Tier: Deckhand Storage 230.5 MB / 5.0 GB progress bar showing 4% used Unique Blobs: 20
Been working on adding quotas to the Holds (storage sidecars)
These are optional and fully configurable with different tiers that can be added to crew records. It is set to 5GB currently for the provided storage on @hold01.atcr.io
The largest account so far is 3.3GB
Image of atcr.io showing Tags. 0.0.1 Helm sha256:.... helm pull oci://atcr.io/evan.jarrett.net/loom --version 0.0.1
New in atcr.io Preliminary support for helm charts.
Shows the lexicons for atcr.io found here https://pdsls.dev/at://did:plc:wfj5kyialpmcv2fzk6uqwsln/com.atproto.lexicon.schema
I haven't looked at the lexicons till tonight for @atcr.io , but I love the naming lol
You can also specify a specific markdown file with
--label io.atcr.readme="https://tangled.org/evan.jarrett.net/at-container-registry/raw/main/docs/appview.md"
Also new in atcr.io
if you add a source label to your build:
--label org.opencontainers.image.source="github or gitlab or tangled url"
The README will be fetched and displayed as the description.
atcr.io showing a pointer finger hovering over an icon that is dimmed out and shows a + that indicates you can edit it.
New in atcr.io you can edit your repo's avatar from the UI.
This requires updated oauth scopes, so you will need to log out and back in.
Note this also applies using labels/annotations when building:
--label io.atcr.io="https://my.cool/image.png"
New feature: open graph images
atcr.io/r/atcr.io/ap...
Screenshot of atcr.io showing a Vulnerability Report modal. 1 High 3 Medium table showing CVE, Severity, Package, Installed, Fixed.
Looking into getting vulnerability scans on atcr.io containers with syft and grype.
Screenshot of atcr.io that showcases darkmode, lucide icons, and embedded readme.
New In atcr.io. Dark mode, and lucide icons.
@atcr.io just pushed hold:latest
Digest: sha256:1ddc828a25... Size: 12.1 MB
@atcr.io just pushed appview:latest
Digest: sha256:8a9052ba4d... Size: 15.0 MB