Halloumi2025h2 and Gouda2025h2

105 views
Skip to first unread message

Pim van Pelt

unread,
Jan 7, 2026, 5:49:03 PMJan 7
to Ct Policy, Ct Ops
Hoi folks,

The write window for these two logs has closed on 2026-01-01 and here's
our plan for these two production logs:

1. I am making a photocopy of them today, Gouda is a Sunlight log and
Halloumi is a TesseraCT log.
2. On Weds 14th, I will:
a) remove the write path for these (down the TesseraCT job, reload the
Sunlight job, removing the logshard entries from their YAML file)
b) Remove upstream *.log.ct.ipng.ch nginx entry for the write path.
c) Publish the ZIP files to our S3 cluster and index them.
d) Update ct.ipng.ch page and publish
https://ct.ipng.ch/archive/halloumi2025h2/ and
https://ct.ipng.ch/archive/gouda2025h2/
e) Send PR for https://github.com/geomys/ct-archive/README.md
3. On Weds January 21st, I will:
a) Remove upstream *.mon.ct.ipng.ch read path nginx entry
b) Remove DNS entries for the log shards (halloumi2025h2.mon.ct.ipng.ch
and halloumi2025h2.log.ct.ipng.ch
4. On Weds January 28th, I will remove the ZFS dataset (zfs destroy
ssd-vol0/logs/halloumi2025h2)

Since this is the first time we're transitioning a completed logshard, I
welcome any feedback or tweaks to our proposed plan.

groet,
Pim (obo IPng CT Ops)
--
Pim van Pelt <p...@ipng.ch>
PBVP1-RIPE https://ipng.ch/

Filippo Valsorda

unread,
Jan 12, 2026, 5:55:02 PMJan 12
to Pim van Pelt, Certificate Transparency Policy, Ct Ops
2026-01-07 23:48 GMT+01:00 'Pim van Pelt' via Certificate Transparency Policy <ct-p...@chromium.org>:
1. I am making a photocopy of them today, [...]
2. On Weds 14th, I will:
a) remove the write path for these [...]

It'd be best to first stop the Sunlight writer for good, and then run photocamera-archiver, so that the archived tree head is the latest one ever signed.

Also, note that the moment that you take the Sunlight shard down, Skylight's /health will get unhappy because it will see a stale checkpoint for a production log. https://github.com/FiloSottile/sunlight/pull/53 fixes this, but I am not recommending you update to a pre-release version without first letting the Geomys logs eat the risk. Instead, you might want to flag that log as Staging in Skylight's config file, so errors will not affect the overall /health HTTP code.
Reply all
Reply to author
Forward
0 new messages