Cluster roster: each node's mail and connections over the period #80

Merged

jcoffey-dev merged 1 commits from feat/node-sparklines into main

2026-10-05 22:13:22 +00:00

jcoffey-dev commented

2026-10-05 22:10:50 +00:00

Owner

Admin UX roadmap item 15, the "sparklines in tables" half. The server keeps history server-wide or per node, so the table it fits is the cluster roster.

What it does

Each node's card in Dashboard › Cluster gets two strips of the period (24 slices), each with its total: Mail (received and sent) and Connections (SMTP, IMAP, POP3, HTTP and ManageSieve connections opened).
Strips follow the period and a chart zoom like everything else on the page; a silent node's strips turn red.
They come from the samples that node wrote (nodeId, servers from 2026.9.30 on). With older servers no sample says which node wrote it, and the strips stay hidden.
A sample's nodeId is the node's cluster id: the server seeds sample ids with SnowflakeIdGenerator::set_node_id(bp.node_id()), the same number x:ClusterNode.nodeId reports.

Also: the strip under a readout is now BarStrip, shared by both.

Strings: 2 new English strings (cc.nodeMail, cc.nodeConnections). No other catalogues yet.

Checked

npm test (61 files, 593 tests; 2 new for nodeSeries), typecheck, lint, i18n check, build.
Not checked in a browser: the roster only shows on a cluster, and the local preview is one node. Worth a look on the cluster's console before or after deploy.

Admin UX roadmap item 15, the "sparklines in tables" half. The server keeps history server-wide or per node, so the table it fits is the cluster roster. **What it does** - Each node's card in Dashboard › Cluster gets two strips of the period (24 slices), each with its total: **Mail** (received and sent) and **Connections** (SMTP, IMAP, POP3, HTTP and ManageSieve connections opened). - Strips follow the period and a chart zoom like everything else on the page; a silent node's strips turn red. - They come from the samples that node wrote (`nodeId`, servers from 2026.9.30 on). With older servers no sample says which node wrote it, and the strips stay hidden. - A sample's `nodeId` is the node's cluster id: the server seeds sample ids with `SnowflakeIdGenerator::set_node_id(bp.node_id())`, the same number `x:ClusterNode.nodeId` reports. **Also:** the strip under a readout is now `BarStrip`, shared by both. **Strings:** 2 new English strings (`cc.nodeMail`, `cc.nodeConnections`). No other catalogues yet. **Checked** - `npm test` (61 files, 593 tests; 2 new for `nodeSeries`), typecheck, lint, i18n check, build. - **Not checked in a browser:** the roster only shows on a cluster, and the local preview is one node. Worth a look on the cluster's console before or after deploy.

jcoffey-dev added 1 commit 2026-10-05 22:10:51 +00:00

Cluster roster: each node's mail and connections over the period

ci / github (pull_request) Skipped

Details

github/ci (branch) GitHub Actions

Details

ci / build (pull_request) Successful in 1m55s

Details

ci / publish (pull_request) Skipped

Details

3a15a2e1be

Admin UX roadmap item 15, "sparklines in tables". Each node's card in
the cluster roster gets two strips of the period, with totals: mail
(received and sent) and connections opened. They come from the samples
that node wrote; servers before 2026.9.30 don't say which node wrote a
sample, and then the strips stay hidden.

The strip of bars under a readout becomes BarStrip, so the roster draws
the same one.

2 new English strings; no other catalogues yet.

jcoffey-dev merged commit dc1fd87749 into main

2026-10-05 22:13:22 +00:00

jcoffey-dev deleted branch feat/node-sparklines

2026-10-05 22:13:22 +00:00

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: inbuxa/inbuxa-admin#80