ref
https://linear.app/ghost/issue/ENG-1746/enable-ghost-to-push-metrics-to-a-pushgateway
- We'd like to use prometheus to expose metrics from Ghost, but the
"standard" approach of having prometheus scrape the `/metrics` endpoint
adds some complexity and additional challenges on Pro.
- A suggested simpler alternative is to use a pushgateway, to have Ghost
_push_ metrics to prometheus, rather than have prometheus scrape the
running instances.
- This PR introduces this functionality behind a configuration.
- It also includes a refactor to the current metrics-server
implementation so all the related code for prometheus is colocated, and
the configuration is a bit more organized. `@tryghost/metrics-server`
has been renamed to `@tryghost/prometheus-metrics`, and it now includes
the metrics server and prometheus-client code itself (including the
pushgateway code)
- To enable the prometheus client alone, `prometheus:enabled` must be
true. This will _not_ enable the metrics server or the pushgateway — it
will essentially collect the metrics, but not do anything with them.
- To enable the metrics server, set `prometheus:metrics_server:enabled`
to true. You can also configure the host and port that the metrics
server should export the `/metrics` endpoint on in the
`prometheus:metrics_server` block.
- To enable the pushgateway, set `prometheus:pushgateway:enabled` to
true. You can also configure the pushgateway's `url`, the `interval` it
should push metrics in (in milliseconds) and the `jobName` in the
`prometheus:pushgateway` block.
- we can skip an extra `toJSON` call here by storing the value and
re-using it, which means we can cut down on the number of calls, which
can be particularly heavy when this codepath is hit hard
refs https://github.com/TryGhost/Ghost/issues/19839
refs https://www.sqlite.org/limits.html
SQLite has limit of 500 items in a compound select statement.
This limit could be hit when a complex select statement was being
generated as part of a batch insert statement.
Lowering the batch size will have minimal impact on migration
performance while improving SQLite compatibility.
One of these bulk inserts is confirmed to be affected through the linked
issue.
I didn't confirm if the other two cases would trigger it, but the change
won't hurt there either.
ref https://linear.app/tryghost/issue/ENG-1556/
- added background job queue behind config flags
- when enabled, is only used for the member email analytics updates in
order to speed up the parent job, and take load off of the main process
that is serving requests
The intent here is to decouple certain code paths from the main process where it is unnecessary, or worse, where it's part of the request. Primary use cases are email analytics (particularly the member stats [open rate]) which are not particularly helpful in the period immediately following an email send, while the click traffic and delivered/opened events are.
Related, the email link clicks themselves send off a cascade of events that are quite a burden on the main process currently and are somewhat tied to the request response when they needn't be. We'll be looking to tackle that after some initial testing with the email analytics job.
ref PLG-220
- Added an `orderAttributes` override method to be able to pass
`count__likes` to the `findPage` DB helper.
- Unknowingly, without that override method in the model, it would strip
all 'non-default' queries.
- Adding that means we could remove our custom database queries and use
the regular `findPage` helper that also handles pagination.
no issue
This change adds the ability to pass a desired language/locale into
portal, as a data-locale tag added to the script that loads it.
Falls back to the previous behavior, loading from the site settings, if
no data-locale is present. (And then to 'en', if necessary.) No change
in expected behavior if data-locale is not passed.
This provides an option to run portal in a language other than the
sitewide language.
(ref: https://app.slack.com/client/T025584C4/C018EKC56JF)
- When we added i18n for search we missed gating it behind the i18n flag.
- There aren't that many translations for search yet, so it's likely not many have noticed yet
- We'll remove the flag soon, but until then, adding the flag for consistency :)
no ref
{{ghost_head}} is huge, and some power-users and theme creators want the
ability to customize what it contains. This PR makes it easier for a
theme to write custom schema, or to load a custom version of
portal/comments/search/etc, or to minimize load times by not loading
scripts where they aren't needed, in a theme-specific way.
Because ghost_head is controlled at the theme level, this gives folks in
managed hosting the new ability to load a different version of the
included app scripts (by preventing ghost_head from writing them and
adding them in manually).
Usage example: ` {{ghost_head exclude="search,portal"}} `
(empty array)
No changes to current behavior
search
The built-in sodo-search script
Includes adding the click event listener on buttons, generating the
search index, and the UI.
portal
The portal script
Handles sign-in and sign-up, payments, tips, memberships, etc, and all
the portal data-attributes.
announcement
The announcement bar javascript
If you'd like to use the announcement bar admin settings but not have it
[mess up your CLS
metric](https://www.spectralwebservices.com/blog/announcement-bar-a-review/),
this is for you.
metadata
Skips HTML tags for meta description, favicon, canonical url, robots,
referrer
Important for SEO
schema
The LD+JSON schema
Important for SEO
card_assets
Loads cards.min.css and .js
Needed on any page with a post body, unless your theme replaces them
all. Assets can also be selectively loaded with the [card_assets
override](https://ghost.org/docs/themes/content/?ref=spectralwebservices.com#editor-cards)
comment_counts
Loads the comment_counts helper
Needed if the page is using {{comments}} or data-ghost-comment-count
attribute
social_data
Produces the og: and twitter: attributes for social media sharing and
previews
Required for good social media cards
cta_styles
Removes the call to action (CTA) styles
Used for member signup and CTA cards - may be overwritten by your theme
already
ref https://linear.app/ghost/issue/ONC-506
- adding `context` with the returned API response makes the logged error much more useful as without it we only log the status code which misses any details for why the failure occurred
no issue
This PR adds the ability to translate the strings that appear in the
newsletter as boilerplate text, using i18next.
Variables are in single mustaches ( `{date}` ) in the translation
strings (rather than `{{date}}`), because these strings occur both the
email template.hbs and also .js files. That necessitated a separate
namespace.
This PR also includes changes to the newsletter button ("more like
this", "less like this", "comment") that were previously delivered on
desktop as images that included the text. @sanne-san provided a rework
that removed text-as-image from the desktop buttons, and allows more
shared code between the two layouts, along with making the buttons
translatable.
Example usage - handlebars
```
<h3 class="latest-posts-header">{{t 'Keep reading'}}</h3>
{{{t 'By {authors}' authors=post.authors }}}
```
(NOTE: triple { required because of possible & )
Example usage - javascript
```
getValue: (member) => {
if (member.status === 'comped') {
return t('complimentary');
}
if (this.isMemberTrialing(member)) {
return t('trialing');
}
// other possible statuses: t('free'), t('paid') //
return t(member.status);
}
```
---------
Co-authored-by: Sanne de Vries <sannedv@protonmail.com>
Co-authored-by: Steve Larson <9larsons@gmail.com>
fix https://linear.app/ghost/issue/ENG-1105/settingscacheget-is-slow
- through profiling and flamegraphs, we can see that `_doGet` is one of
the bottlenecks during high traffic times, sometimes taking up to 20%
of the CPU time when hammering Ghost with `wrk`
- this is because, for the majority of settings cache lookup, we're
running `JSON.parse`, which blocks the main thread
- whilst we're only parsing small strings, we're doing it a LOT,
sometimes hundreds of times per request, which adds up
- this code just throws most deserializing at `JSON.parse`, so if we can
stop it from doing that, it'd be a huge win
- my initial attempts here were to convert the _doGet function to a
smarter deserializing, by looking up `cacheEntry.type` and acting
accordingly
- however, it became a bit of a logical nightmare, and difficult to
reason about for now (i still think we should do it)
- therefore, I'm just doing to add a hotpath fix to catch 99% of
usecases, which is checking the type of the cache entry and returning
the value if it's a string
- on a trivial benchmark locally, this causes Ghost to return 30% more
requests per second!!
ref PLG-220
- Improved `getBestComments` service to paginate correctly since we're
using a custom query to determine the top comments that goes beyond the
scope of what `findPage` is capable of.
- Updated CommentsController and CommentsService to support custom order
parameters.
- Added tests
This adds a `content_api_url` helper, returning the url for Ghost's
Content API. By default it will return an absolute URL but can be
passed `absolute=false` if a relative URL is wanted.
This works in tandem with the `content_api_key` helper to
facilitate third party integrations with the Content API, for
example - custom Portal or Search implementations.
no issue
- Fixed a flaky [publishing
test](https://github.com/TryGhost/Ghost/actions/runs/11509561903/job/32039943951)
that was suffering from a race condition. It was trying to copy the
bookmark link shown on the publishing complete modal, but it was
sometimes already closed by that point.
- It seemed to pass consistently locally, but in CI it would frequently
fail. This commit should wait to copy the link before closing the modal.
no issue
- Dev Containers let you work on Ghost in a consistent, isolated
environment with all the necessary development dependencies
pre-installed. VSCode (or Cursor) can effectively run _inside_ the
container, providing a local quality development environment while
working in a well-defined, isolated environment.
- For now the default setup only works with "Clone repository in
Container Volume" or "Clone PR in Container Volume" — this allows for a
super quick and simple setup. We can also introduce another
configuration to allow opening an existing local checkout in a Dev
Container, but that's not quite ready yet.
- This PR also added the `yarn clean:hard` command which: deletes all
node_modules, cleans the yarn cache, and cleans the NX cache. This will
be necessary for opening a local checkout in a Dev Container.
- To learn more about Dev Containers, read this guide from VSCode:
https://code.visualstudio.com/docs/devcontainers/containers#_personalizing-with-dotfile-repositories
---------
Co-authored-by: Joe Grigg <joe@ghost.org>
Co-authored-by: Steve Larson <9larsons@gmail.com>
- Adding custom fonts for themes behind a feature flag
- Introduces new `@tryghost/custom-fonts` module to manage custom fonts
- UI updates for Branding and Theme settings
---------
Co-authored-by: Fabien O'Carroll <fabien@allou.is>
Co-authored-by: Sodbileg Gansukh <sodbileg.gansukh@gmail.com>
Co-authored-by: Peter Zimon <peter.zimon@gmail.com>
Co-authored-by: Sanne de Vries <sannedv@protonmail.com>
Co-authored-by: Daniël van der Winden <danielvanderwinden@ghost.org>
- replaced a couple of uses of lodash.each in favor of native for loops
- tidied up `debug` statements and spacing
- pulled out common statements into variables
closes https://linear.app/ghost/issue/ONC-254closes#20771
The portal script, which is responsible for handling the recommendations popup,
was only loaded into the front end if either members or donations are enabled.
We're adding an extra condition to load it if recommendations are enabled.
We may want to consider splitting out this functionality into several scripts,
so that we don't have to load _everything_ if only one feature is enabled, but
that is outside the scope of this issue.
refs https://linear.app/ghost/issue/ONC-469
Hidden comments were not being purged from the cache, which resulted
in stale data being served, and hidden comments being visible.
refs https://linear.app/ghost/issue/ONC-469
Hidden comments were not being purged from the cache, which resulted
in stale data being served, and hidden comments being visible.
