You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The tlog-tiles and static-ct-api specs allow partial tiles to be deleted
when the corresponding full tile is available. This helps to reduce R2
storage costs, but will incur extra cost for the R2 list and delete
operations.
- Add a `Cleaner` Durable Object, which iterates over a log and for each
full tile available in the public bucket, lists and deletes any
corresponding partial tiles. The cleaner tracks subrequests to prevent
an alarm from exceeding the 1000 subrequests limit per invocation.
- (bonus) Lays the groundwork for implementing a tlog-witness or
tlog-mirror as a service that periodically updates based on a target
log's latest checkpoint.
Other changes:
- Update to worker 0.6.1 for R2 `delete_multiple` support and
`PartialEq` support for ObjectIds.
- Move more Durable Object intialization logic from the MTC and CT
applications to the `generic_log_worker` crate to deduplicate code.
- Fix tree time metric to seconds.
Copy file name to clipboardExpand all lines: crates/ct_worker/config.schema.json
+6Lines changed: 6 additions & 0 deletions
Original file line number
Diff line number
Diff line change
@@ -113,6 +113,12 @@
113
113
"type": "boolean",
114
114
"default": true,
115
115
"description": "Enables loading root store trusted roots from the CCADB list, in addition to any roots configured in `roots.<env>.pem`. If enabled, requires a KV namespace with the binding `ccadb_roots` to be configured in `wrangler.jsonc`, as well as a cron trigger so that the CCADB list auto-updates."
116
+
},
117
+
"clean_interval_secs": {
118
+
"type": "integer",
119
+
"minimum": 1,
120
+
"default": 60,
121
+
"description": "How long to wait in between runs of the partial tile cleaner. For static CT, the cleaner can clean 498 tiles (127,488 entries) per run before hitting the Workers limit of 1000 subrequests, so the default of once every 60 seconds should keep up with a log that grows at 2000 entries/second."
0 commit comments