Stephen Wicklund issues

Results 189 issues of


                                            Stephen Wicklund

Get metrics-server working for Lens and such (at least for some key metrics such as node storage-space-left, as hit disk-pressure issue that's how to anticipate otherwise)

The temporary fix, when disk-pressure occurs: * Run: `docker system prune`

Improve tooling around access-policies, to make it similar to using a map-level permission-setting (for typical use-cases)

backend

frontend

Disable auto-backups of Firestore instance (since no longer in use), and clean up existing stored backups (the auto-backups have amassed significant storage usage)

Complete conversion of rest of JSONValue fields in db-structs that can be parsed as actual structs

Examples: * `NodeRevision.displayDetails` * `Attachment.equation`, `Attachment.XXX`

backend

Add way for user to list node-ratings, etc. that are attached to entries they can no longer access, enabling them then to delete those derivative-entries

Benefit: Lets say someone publicly rates NodeX; the creator of NodeX then ends up making the node private; the person who rated NodeX changes their mind and wants to remove...

Make-so there is some sort of local "backup" system in place for prod-builds (for the frontend js bundles), to enable easy publish-reverts

Possibilities: * Each time the Dist folder is changed, create a local backup, using a halving-to-past schedule. * Just make it easy to find/apply old docker images in the GCP...

Improve the Logs->Stored panel in monitor-tool (eg. by having the Groups settings apply to it, at least somewhat)

Maybe add logic to monitor-backend that logs errors (to then trigger sentry email) if k8s issues are noticed in cluster (eg. pod crashes, disc running out, etc.)

backend

Improve security of cross-namespace kubernetes permission-grants, by consolidating that logic into one custom "operator" pod (safer, since its api will be small, and easier to audit; also, can replace the finicky/unreliable "reflector" pod)

backend

MS errors that occur on frontend do not show the stack-trace in the notification (or at least, only show it if the user specifically expands it, or presses "Copy debug info" or something -- or if user sets pref to always show stack-trace [nice for devs])

good first issue

frontend