[META] Synthetics Beta Release

[andrewvc]

Decision items

• Syntax finalized
• Schema finalized
• Decision on browser (e.g. Chromium) and framework (e.g. Playwright)
• Bundle Synthetics Agent with Heartbeat Docker container [Heartbeat] Add synthetics deps to main heartbeat docker image beats#22837

Other

• Implement more mature packaging strategy Synthetics Beta Packaging, Docker, Fleet, Build, and other Concerns #140
• Merge synthetics feature branches into beats mainline Initial work toward synthetic monitors beats#21436
• Add E2E / Functional tests between Heartbeat <-> Synthetics <-> UI Add E2E Test Suite / Example #115
• Add Functional test for CLI Add Functional test for CLI #139
• Unit tests for network/waterfall timing calculation Unit tests for network/waterfall timing calculation #185
• Migrate to regular heartbeat docker image Migrate examples to use heartbeat snapshot docker containers instead of experimental synthetics ones #182
• Complete docs
  • Docs/release note to make it clear that the latest version of the Synthetic Agent is needed (specifically, to support the waterfall chart)
  • Document package.json for synthetics Need to discuss package.json in synthetics docs observability-docs#284
  • Document how to use/deploy Docker images (not run.sh) Document how to use/deploy Docker images (not run.sh) observability-docs#300
• Update notices and banners from experimental to beta [Uptime] Update notices and banners from experimental to beta kibana#91421
• Run Playwright post-install script to fix chrome binary sometimes unavailable Run Playwright post-install script to fix chrome binary sometimes unavailable #142
• Feature flag to control filmstrip image capture Feature flag to control filmstrip image capture #143
• (Bug) Docker example doesn't work well with uids != 1000 Docker example doesn't work well with uids != 1000 #156
• Redirects and DNS failures not represented correctly Redirects and DNS failures not represented correctly #175
• no-sandbox option
  • Synthetics implementation Add option to disable chromium sandbox #136
  • Heartbeat implementation [Heartbeat] Add no_chromium_sandbox option for synthetics beats#22901
• Release Synthetics as package version 1.0.0-beta at FF (early enough for BCs)
• Move network info fields to ECS Move network fields to ECS where possible (starting with network data) #164
• Move timing calculation to agent capture more network information to improve waterfall #152
• Exit with non-zero code when any step fails exit with non-zero code when any step fails  #191
• Source is transpiled to lower JS version for suites Transpiled source is displayed instead of raw typescript #172
• Plan for handling synthetics dependencies [Heartbeat] Plan for handling synthetics dependencies beats#22928
• Remote Zip URL support for heartbeat synthetics Remote Zip URL support for heartbeat synthetics uptime#271

[drewpost]

@andrewvc - I would like to support at least Webkit in addition to Chromium in a production product so can "decision on browser" be scoped to look at having two browsers in the Playwright package?

[andrewvc]

@drewpost going from 1 -> 2 browsers requires investigation in terms of handling perf data, we use the chrome debug protocol to extract perf data now. Supporting web kit would mean normalizing two formats AFAIK

@jahtalab @vigneshshanmugam your thoughts?

CC @paulb-elastic

[hmdhk]

At the moment, a lot of key features depend on chrome devtools protocol (e.g. Tracing, Network, Performance) which means these won't be available for other browsers if we choose to add them.

So I see two options:

1. We add multi-browser support but with the understanding that those features won't be available. This is lower effort, but the UI needs to account for the missing features (and UX will suffer)
2. We can investigate ad-hoc solutions to bring some of those features to other browsers. This is high effort and there's no guarantee that we can bring all of the features.

I'm leaning towards option 1. but we should make sure that the set of features available can be at least somewhat useful for the users.

[paulb-elastic]

We may also want to consider focusing on a single browser for now (in beta and even in GA), as we build up features and capabilities.

As soon as we add another browser, it at least doubles the testing we need to do, plus the additional complexity of any code divergence to handle both and understanding which features work in which browsers (which could become difficult to track).

[vigneshshanmugam]

There is an easy way out here, We can support webkit by calling out as some features as minimal pointed already in the thread. Minimal here means

• We can already get the Waterfall working as Playwright already supports way of extracting Request/Response data.
• Filmstrips wont be possible at the same sampling interval as chromium as we rely on Chrome Devtools protocol. However, we can support filmstrips by capturing screenshots which would make out tests slower but still possible
• Performance metrics, we don't surface it at the moment so not a huge concern.

But personally for the Beta, I would still prefer to keep it to one browser and understand our customer base and then go full fledged.

[drewpost]

Happy to keep it one browser for beta as long as there's scope for adding in another in the future.

[andrewvc]

OK, it sounds like the decision here is to keep to one browser and revisit in the future then. Is that right?

[paulb-elastic]

Some items moved from beta to the GA phase

[paulb-elastic]

Categorised the requirements into those that are awaiting a decision, and those that require an implementation

[paulb-elastic]

Added the requirement to change the behaviour such that the Synthetic Agent does not capture filmstrip images by default, but enable them via a feature flag/configuration (#143)

[paulb-elastic]

Modified the link to refer to https://github.com/elastic/uptime-dev/issues/32 which replaces https://github.com/elastic/uptime-dev/issues/31

[paulb-elastic]

Moved #144 from beta to GA deliverable

[paulb-elastic]

As per this discussion, we are moving the bigger security changes https://github.com/elastic/uptime-dev/issues/32 from beta to GA, but adding the sandbox option (synthetics implementation, heartbeat implementation) as a beta requirement.

[paulb-elastic]

Added #185 to the scope of beta

[paulb-elastic]

Added #191 as a beta requirement

[vigneshshanmugam]

Adding #172 as a beta requirement as its a bug and happens for all suites that are run as TS files.

[paulb-elastic]

Moving remote ZIP support from being in GA to be in for beta

[paulb-elastic]

The above change of scope has been reverted. Remote Zip URL support for heartbeat synthetics has been taken out of this (beta) and added back into GA.

[paulb-elastic]

elastic/uptime#271 made beta, so added back into here

[vigneshshanmugam]

package.json docs are here - https://discuss.elastic.co/t/synthetics-1-0-0-beta-0-breaking-changes/271118