waitUntil called on event that is already finished in NetworkFirst strategy

[jeffposnick]

Library Affected:
workbox-strategies

Browser & Platform:
webview with User-Agent:Mozilla/5.0 (Linux; Android 6.0.1; SM-A510F Build/MMB29K; wv) AppleWebKit/537.36 (KHTML, like Gecko) Version/4.0 Chrome/56.0.2924.87 Mobile Safari/537.36

Issue or Feature Request Description:
Reported by @beatrizdemiguelperez on Slack:

i'm getting this error while trying to fetch a file with networkFirst strategy

Uncaught (in promise) DOMException: Failed to execute 'waitUntil' on 'ExtendableEvent': The event handler is already finished.
   at http://localhost:4200/workbox-v3.0.0/workbox-strategies.dev.js:510:15
   at Generator.next (<anonymous>)
   at step (http://localhost:4200/workbox-v3.0.0/workbox-core.dev.js:14:32)
   at http://localhost:4200/workbox-v3.0.0/workbox-core.dev.js:25:15

[jeffposnick]

The underlying code is

workbox/packages/workbox-strategies/NetworkFirst.mjs

Lines 220 to 229 in 49f3d21

	// Keep the service worker alive while we put the request in the cache
	const responseClone = response.clone();
	event.waitUntil(
	cacheWrapper.put(
	this._cacheName,
	event.request,
	responseClone,
	this._plugins
	)
	);

There's some setTimeout() logic in the handler that might be leading to a race condition.

[beatrizdemiguelperez]

Also reproduced on:
Browser & Platform:
Chrome Mobile with User-Agent:Mozilla/5.0 (Linux; Android 7.1.1; SM-N950F Build/NMF26X) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.125 Mobile Safari/537.36
Issue or Feature Request Description:
It happens after refreshing page and trying to fetch a file. This file was fetched before refreshing and it has networkFirst strategy

[mick-feller]

We have been noticing the same issue happening.

It's hardly impossible to reproduce but we got it on a handful of machines, what happened (or at least i think) we had some backend issue's and the service workers kicked in during this time. At that moment something is going south and i'm getting this exact same issue. I can confirm that on version 3.0.1 it was stopping the service worker all together and serving empty files. I've updated to version 3.2.0 and at least the service worker is now returning content.

All though it seems impossible to recover the service worker after this has happened unless you forcefully unregister the service worker and reinitialize it.

My caching strategy is mixed between staleWhileRevalidate, cacheFirst and preCache.

I wish i could put a case together were I could 100% replicate it but it's near impossible at this moment.

Do you have any pointers for me in how to recover the service worker after something like this happens?

Thanks