Idea for dev server - simulated infrastructure instability

[tempy]

Hi all,

I just found myself writing code that would simulate deadlineexceeded
errors and such, when I thought - wouldn't it be great if you could
ask the dev server to sow some chaos? I'll never be able to accurately
simulate outages, slow-downs, high-error rates, variable accuracy of
the infrastructure-availability API, and all the possible varieties of
misbehavior from the GAE infrastructure. But Google probably can.

What I'm imagining is asking the dev server to execute certain
"degraded service" scenarios. So you could ask it to simulate
"memcache instability" which would result in an unusually high number
of cache misses, slow cache responses, and the occasional exception.
Or you could ask it to simulate a "read only datastore" event. And so
on - I think you guys get my point.

Anyone else think this would be incredibly useful?

-mike

[Strom]

I don't think just having the memcache not work on the dev server
would have much of an impact, because the dev datastore is fast
anyway. However I definitely see a use for a "slow mode" where API
calls take around the same time as they do in production, including
some occasional crazy delays.

[Robert Schuppenies]

I think it's a good idea. Can you file a feature request for this?

--
You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To post to this group, send email to google-a...@googlegroups.com.
To unsubscribe from this group, send email to google-appengi...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.

[tempy]

Done:

http://code.google.com/p/googleappengine/issues/detail?id=7453

[Robert Schuppenies]

Thanks!

[Dale]

I've added an extra note to the feature request for running the same task multiple times.