refs: https://github.com/TryGhost/Toolbox/issues/146
Switched to @tryghost/logging instead of passing around the library. The main sticking points of this change are jobs. When jobs are launched we don't want them to use a separate @tryghost/logging instance because they would start parallel rotation jobs. @tryghost/logging v2.x passes all logs to the parent process if run in a child process, so that we can use the same patterns in jobs and the rest of the codebase.
no issue
- `startsWith` method is way easier to read and understand. also, **probably** has better performance comparing to building up a regexp and then matching
- This is part of the quest to separate the frontend and server & get rid of all the places where there are cross-requires
- At the moment the settings cache is one big shared cache used by the frontend and server liberally
- This change doesn't really solve the fundamental problems, as we still depend on events, and requires from inside frontend
- However it allows us to control the misuse slightly better by getting rid of restricted requires and turning on that eslint ruleset
fixes https://github.com/TryGhost/Team/issues/834
- see referenced issue for context
- there are times when `parentPort` can be null and the job crashes
because `parentPort.postMessage` won't work
- this commit adds guards around `parentPort`, or moves code inside
existing guards, to protect against this
no issue
The only pieces of Ghost-Ignition used in Ghost were debug and
logging. Both of these modules have been superceded by the Framework
monorepo, and all usages of Ignition have now been removed, replaced
with @tryghost/debug and @tryghost/logging.
refs https://github.com/TryGhost/Ghost/pull/12541
refs https://github.com/TryGhost/Ghost/pull/12689
- the analytics job had been switched to create it's own instance of EmailAnalyticsService to avoid requiring logging but the analytics extraction branch was created before this change and wasn't picked up when merging
- pulled `queries` option object into a separate file for re-use
- updated `fetchLatest` job to conform to extracted library interface
closes https://github.com/TryGhost/Team/issues/493
- all functionality except that directly related to Ghost's database and business logic now lives in external packages
- @tryghost/email-analytics-service
- @tryghost/email-analytics-provider-mailgun
refs https://github.com/TryGhost/Team/issues/221#issuecomment-759105424
- Mailgun responds to an email send with a provider id in the format `<x@y.com>` but everywhere else it's used in their API it uses the format `x@y.com`
- updates email batch save to strip the brackets, and migration removes brackets from existing records so we no longer have to add special handling for the stored id any time we use it
refs https://github.com/TryGhost/Ghost/issues/12496
By requiring the models layer the shared logging util was being required as a side-effect causing the open file descriptors problem to continue. Removing logging from the models layer isn't feasible due to deep require chains spreading across the codebase, it's much quicker to remove the need for models in the analytics job.
- models layer was only needed because it's used by the session service
- updated analytics job to create it's own instance of `EmailAnalyticsService` rather than the default instance in order to pass in custom dependencies
- pass in custom `logging` object that uses `parentPort.postMessage` as a way of writing log output
- pass in custom `settings` object that returns settings that have been manually fetched and cached during job instantiation
refs https://github.com/TryGhost/Ghost/issues/12496
- Using ghost-ignition logging caused file handle leaks. As there is no straight-forward way to handle write streams with bunyan (ghost-ignition's underlying logging library) this method of logging was chosen as an alternative to keep the amount of open file handles to minimum
- The follow up changes will include custom formatter for jobs service which should make logging match the same format as has been used inside the jobs
refs https://github.com/TryGhost/Team/issues/446
Mailgun permanent failure events do not always correspond to unsubscribe-level events as originally thought, meaning some members could be unsubscribed unexpectedly due to delivery hiccups.
- disabled auto-unsubscribe on permanent failure events in the analytics event processor
- list maintenance will be added back in the future via alternative means
refs https://github.com/TryGhost/Ghost-Utils/issues/118
- Duplicating error handling across jobs is not best developer experience. Also, having custom error handling logic did not allow for recommended worker script behavior: allowing for unhandled exceptions to bubble up and be managed by parent process
no issue
- sqlite will store a float in an integer column due to it's type affinity resulting in long decimal numbers in the UI when we're expecting an integer
- use the `ROUND()` function to ensure we're storing integers in place of floats when performing open rate average calculations
refs https://github.com/TryGhost/Ghost/issues/12461
- added two default aggregations for overall email count and opened email count
- when number of tracked emails is sufficient add the open rate aggregation to the update query
refs https://github.com/TryGhost/Ghost/issues/12421
requires https://github.com/TryGhost/Ghost/pull/12457
- updates stats aggregator to calculate and store an open rate for each member
- uses two queries because I couldn't find a reasonable approach to perform the update in a single query as per the email aggregation
- benchmarked locally at <1sec/1000members
- will not store an open rate unless the number of tracked emails sent to a member is above a certain threshold (defaults to 5) to avoid new members being heavily weighted
- fixes typo in EmailAnalytics that was stopping member stats from being aggregated
no issue
- job registration was checking for submitted emails in it's email count but the job registration method is called as soon as an email is created meaning the email has a status of 'pending' which prevented the analytics job from being started until a second email was sent
no issue
- email analytics may be desirable to fully switch off in certain circumstances, when that happens we want to prevent related background jobs from running and expose the feature flag via the config endpoint in the Admin API so that clients can adjust accordingly
no issue
- if emails are older than 30 days we wouldn't be able to fetch any analytics for them and if a site used emails in the past but is no longer using them it doesn't make sense to keep potentially expensive background worker threads spinning up
no issue
- recurring jobs spin up worker threads which can be quite CPU intensive even when not performing much processing, this can be problematic in environments where there are many Ghost instances running
- updated the email job scheduling to be skipped on bootup when there are no emails in the database and to be started when the first email is created as long as we're not in testing env
- increase analytics job schedule from every 2 minutes to every 5 minutes to help spread the load further across instances
no issue
- the 4.2.0 version of `sqlite3` that we're using is not compatible with `worker_threads`
- 5.0.0 should add support this but there are other errors
- 5.0.1 is released but not published (https://github.com/mapbox/node-sqlite3/issues/1386)
no issue
- added `EmailAnalyticsService`
- `.fetchAll()` grabs and processes all available events
- `.fetchLatest()` grabs and processes all events since the last seen event timestamp
- `EventProcessor` passed event objects and updates `email_recipients` or `members` records depending on the event being analytics or list hygiene
- always returns a `EventProcessingResult` instance so that progress can be tracked and merged across individual events, batches (pages of events), and total runs
- adds email_id and member_id to the returned result where appropriate so that the stats aggregator can limit processing to data that has changed
- sets `email_recipients.{delivered_at, opened_at, failed_at}` for analytics events
- sets `members.subscribed = false` for permanent failure/unsubscribed/complained list hygiene events
- `StatsAggregator` takes an `EventProcessingResult`-like object containing arrays of email ids and member ids on which to aggregate statistics.
- jobs for `fetch-latest` and `fetch-all` ready for use with the JobsService
- added `initialiseRecurringJobs()` function to Ghost bootup procedure that schedules the email analytics "fetch latest" job to run every minute
