Closes K-L2 (#9): lab/chart/mailcheck.sh runs every client check against an installed chart.
lab/chart/mailcheck.py, as [email protected], each check named, one table, exit 1 on any failure:
Check
What
smtp in
a message from another domain on 25
submission
alice on 465, implicit TLS and AUTH: to herself, and out to [email protected]
imap
both arrive, found over IMAPS (any mailbox: the inbound one lands in Junk)
jmap
both found with Email/query, signed in with OAuth (JMAP doesn't take Basic)
smtp out
the outgoing one reaches the outside server from alice, DKIM-signed d=lab.test
caldav
an event PUT, read back, DELETEd, then 404
carddav
a contact the same
Then signin.py (the webmail across both replicas) and console-signin.py, as before.
The outside server:lab/chart/mail-sink.yaml, a small SMTP sink (plain Python) in its own namespace, in the CloudNativePG PostgreSQL image the cluster already has, so no new image. It logs each message as a JSON line, which the check reads. Mail for sink.test gets there by a Relay route lab-sink and one line in the outbound strategy, which the check adds as admin if they're missing. So it works on any seeded lab, not just a fresh one.
e2e.sh now runs mailcheck.sh after the install, in place of its own three steps. Plus a one-line fix in tools.sh: kubectl version | head -1 could die of SIGPIPE under pipefail (it failed one CI job).
Tested: the bench lab (all checks pass), and all six CI jobs from scratch (3 Kubernetes minors × amd64/arm64). The checks take about 12 s in all; e2e is ~240 s.
Closes K-L2 (#9): **`lab/chart/mailcheck.sh`** runs every client check against an installed chart.
`lab/chart/mailcheck.py`, as [email protected], each check named, one table, exit 1 on any failure:
| Check | What |
|---|---|
| smtp in | a message from another domain on 25 |
| submission | alice on 465, implicit TLS and AUTH: to herself, and out to [email protected] |
| imap | both arrive, found over IMAPS (any mailbox: the inbound one lands in Junk) |
| jmap | both found with Email/query, signed in with OAuth (JMAP doesn't take Basic) |
| smtp out | the outgoing one reaches the outside server **from alice, DKIM-signed `d=lab.test`** |
| caldav | an event PUT, read back, DELETEd, then 404 |
| carddav | a contact the same |
Then `signin.py` (the webmail across both replicas) and `console-signin.py`, as before.
**The outside server:** `lab/chart/mail-sink.yaml`, a small SMTP sink (plain Python) in its own namespace, in the CloudNativePG PostgreSQL image the cluster already has, so no new image. It logs each message as a JSON line, which the check reads. Mail for `sink.test` gets there by a Relay route `lab-sink` and one line in the outbound strategy, which the check adds as admin if they're missing. So it works on any seeded lab, not just a fresh one.
`e2e.sh` now runs `mailcheck.sh` after the install, in place of its own three steps. Plus a one-line fix in `tools.sh`: `kubectl version | head -1` could die of SIGPIPE under pipefail (it failed one CI job).
**Tested:** the bench lab (all checks pass), and [all six CI jobs](https://github.com/inbuxa/inbuxa-k8s-lab/actions/runs/44687987705) from scratch (3 Kubernetes minors × amd64/arm64). The checks take about 12 s in all; e2e is ~240 s.
lab/chart/mailcheck.py checks, as alice: SMTP in on 25, submission on 465
with AUTH, IMAPS, JMAP (OAuth, Email/query), SMTP out to an outside
server, CalDAV and CardDAV round trips; one table, exit 1 on any failure.
The outside server is lab/chart/mail-sink.yaml, a small SMTP sink in its
own namespace, in the CloudNativePG image the cluster already has; mail
for sink.test reaches it by a Relay route the check sets up as admin. The
outgoing message has to arrive from alice and DKIM-signed for lab.test.
lab/chart/mailcheck.sh runs it with the webmail (across both replicas)
and console sign-ins against an installed chart; e2e.sh now runs that
after the install.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Closes K-L2 (#9):
lab/chart/mailcheck.shruns every client check against an installed chart.lab/chart/mailcheck.py, as [email protected], each check named, one table, exit 1 on any failure:d=lab.testThen
signin.py(the webmail across both replicas) andconsole-signin.py, as before.The outside server:
lab/chart/mail-sink.yaml, a small SMTP sink (plain Python) in its own namespace, in the CloudNativePG PostgreSQL image the cluster already has, so no new image. It logs each message as a JSON line, which the check reads. Mail forsink.testgets there by a Relay routelab-sinkand one line in the outbound strategy, which the check adds as admin if they're missing. So it works on any seeded lab, not just a fresh one.e2e.shnow runsmailcheck.shafter the install, in place of its own three steps. Plus a one-line fix intools.sh:kubectl version | head -1could die of SIGPIPE under pipefail (it failed one CI job).Tested: the bench lab (all checks pass), and all six CI jobs from scratch (3 Kubernetes minors × amd64/arm64). The checks take about 12 s in all; e2e is ~240 s.