Overview - Ghost

7 days30 days

Latest CI Pipeline Executions

Fuzzy

Succeeded
30690
15a66bc1 Changed Mailgun polling to keep progress through cooldowns and shutdown ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api A provider cooldown that outlasted the wait budget failed the whole run and rewound its cursor to the window start, so pages already processed were fetched again next cycle. The Logs adapter now stops such a run as throttled: it keeps what each domain covered and leaves the whole window for domains it never read. Cooldown hints are bounded to one hour so one bad header cannot park every reader until restart, readers waking from a shared cooldown add jitter so they do not fire together, and a reset sent in seconds is read instead of ignored. A planned shutdown is no longer reported as a job failure or a failed provider request, the running flag is cleared when a stop interrupts a run between lanes, reinitializing while interrupted fetches drain warns instead of failing boot, lifecycle tasks register once per server, and a non-boolean prefetch setting is rejected instead of silently ignored.
by Jonatan Sv...
J
Succeeded
30690
a63a3c74 Changed Mailgun polling to keep progress through cooldowns and shutdown ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api A provider cooldown that outlasted the wait budget failed the whole run and rewound its cursor to the window start, so pages already processed were fetched again next cycle. The Logs adapter now stops such a run as throttled: it keeps what each domain covered and leaves the whole window for domains it never read. Cooldown hints are bounded to one hour so one bad header cannot park every reader until restart, readers waking from a shared cooldown add jitter so they do not fire together, and a reset sent in seconds is read instead of ignored. A planned shutdown is no longer reported as a job failure or a failed provider request, the running flag is cleared when a stop interrupts a run between lanes, reinitializing while interrupted fetches drain warns instead of failing boot, lifecycle tasks register once per server, and a non-boolean prefetch setting is rejected instead of silently ignored.
by Jonatan Sv...
J
Succeeded
30690
5fd980b8 Changed Mailgun polling to keep progress through cooldowns and shutdown ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api A provider cooldown that outlasted the wait budget failed the whole run and rewound its cursor to the window start, so pages already processed were fetched again next cycle. The Logs adapter now stops such a run as throttled: it keeps what each domain covered and leaves the whole window for domains it never read. Cooldown hints are bounded to one hour so one bad header cannot park every reader until restart, readers waking from a shared cooldown add jitter so they do not fire together, and a reset sent in seconds is read instead of ignored. A planned shutdown is no longer reported as a job failure or a failed provider request, the running flag is cleared when a stop interrupts a run between lanes, reinitializing while interrupted fetches drain warns instead of failing boot, lifecycle tasks register once per server, and a non-boolean prefetch setting is rejected instead of silently ignored.
by Jonatan Sv...
J
Succeeded
30690
1a63edcd Changed Mailgun polling to keep progress through cooldowns and shutdown ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api A provider cooldown that outlasted the wait budget failed the whole run and rewound its cursor to the window start, so pages already processed were fetched again next cycle. The Logs adapter now stops such a run as throttled: it keeps what each domain covered and leaves the whole window for domains it never read. Cooldown hints are bounded to one hour so one bad header cannot park every reader until restart, readers waking from a shared cooldown add jitter so they do not fire together, and a reset sent in seconds is read instead of ignored. A planned shutdown is no longer reported as a job failure or a failed provider request, the running flag is cleared when a stop interrupts a run between lanes, reinitializing while interrupted fetches drain warns instead of failing boot, lifecycle tasks register once per server, and a non-boolean prefetch setting is rejected instead of silently ignored.
by Jonatan Sv...
J
Succeeded
30690
056ae8d0 Changed Mailgun polling to keep progress through cooldowns and shutdown ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api A provider cooldown that outlasted the wait budget failed the whole run and rewound its cursor to the window start, so pages already processed were fetched again next cycle. The Logs adapter now stops such a run as throttled: it keeps what each domain covered and leaves the whole window for domains it never read. Cooldown hints are bounded to one hour so one bad header cannot park every reader until restart, readers waking from a shared cooldown add jitter so they do not fire together, and a reset sent in seconds is read instead of ignored. A planned shutdown is no longer reported as a job failure or a failed provider request, the running flag is cleared when a stop interrupts a run between lanes, reinitializing while interrupted fetches drain warns instead of failing boot, lifecycle tasks register once per server, and a non-boolean prefetch setting is rejected instead of silently ignored.
by Jonatan Sv...
J
Succeeded
30690
4d53c908 Added analytics polling cooldown and shutdown lifecycle ref https://linear.app/ghost/issue/BER-3936/raise-mailgun-event-fetch-throughput-concurrent-paging-logs-api Keep provider cooldowns across polling cycles and drain interrupted ingestion safely before shutdown, retaining cursors and scheduled backfills for replay.
by Jonatan Sv...
J