aleph

2016-10-07T11:30:30.000316Z

would love to have something like that that integrates with existing dashboard tools (like Kibana)

2016-10-07T15:47:22.000317Z

My plan is to make something that is extensible (via subscribing to streams), so i think that should be possible. I already found that the hystrix metrics granularity is not good enough for us so I’m thinking of other visualizations too. But the hystrix dashboard is a start 🙂

2016-10-07T15:49:25.000318Z

One thing I would really like is to have a visualisation of latent requests. This visualisation should be able to tell what caused a timeout for instance. This visualisation could be a timeline of calls to nested services

dm3 2016-10-07T15:50:19.000319Z

have you seen the latest Zach talk at strange loop?

2016-10-07T15:50:30.000320Z

not sure

2016-10-07T15:50:33.000321Z

i don’t think so

dm3 2016-10-07T15:51:01.000322Z

I think they did the same at Fitbit by having an explicit state machine (passport) passed through the code

2016-10-07T15:51:21.000323Z

sounds interesting

2016-10-07T15:51:27.000324Z

is it on youtube already?

dm3 2016-10-07T15:51:30.000325Z

yeah

2016-10-07T15:51:40.000326Z

cool, i’ll have a look

2016-10-07T15:54:24.000327Z

found it https://www.youtube.com/watch?v=_1rh_s1WmRA

2016-10-07T15:57:46.000329Z

Had a quick glance through it. It has definitely overlap 🙂

2016-10-07T17:08:53.000330Z

@jeroenvandijk: I think we're doing something similar. basically all http requests are tagged, recursively, which we can then use for debugging

2016-10-07T17:09:34.000331Z

so when we know, say, a video upload failed, we can immediately get all the logs related to that initial upload request