We've found NATS to be unstable under certain conditions, temporary network interruptions or network instability, around the client reconnection logic. We've seen that it could take anywhere from a

It frequently logs the message below. It seems not

Please check the file collector/collector.log, it's in a subdirectory of the unpacked log tarball. wrote:

Actually, I checked the two files in status_z1 job VM. I did not find any clues. Attached for reference.

If you do "bosh logs stats_z1 0 --job" you will get a tarball of all the logs for the relevant processes running on the stats_z1/0 VM. You will likely find some error messages in the collectors

Thanks Mohamed and Max for sharing this great work. Besides the supporting an official TCK, the cf-swagger repo seems great to ease the delivery of acceptance tests as part of a a service broker

It does not help. I always see the "collector" process bouncing between "running" and "does not exit" when I use "monit summary" in a while loop. Who knows how to get the real error when the

My approach is to login on the stats vm and sudo, then run "monit status" and restart the failed processes or simply restart all processes by running "monit restart all" wait for a while(5~10 minutes

Hi all, I am installing cf 212 with stemcell 2989. However, I got the failure on status_z1. Started preparing configuration > Binding configuration. Done (00:00:04) Started updating job

Thanks for the announcement. To be clear is this announcement to cease support for the old v1 brokers or is this to eliminate support for the v1 api in the CC? Does the v1 CC code depend on NATS?

If you are sharing a vip for http and websocket then 443 would be correct. But Anthony, you can try creating a layer 4 virtual server on 4443 that goes to the same pool on the back end and configure

Are you sure your logregator endpoint is configured on 443 and not 4443?

Yep. --recent works. Other cf commands and cf curl also works. It definitely is the websockets for loggregator. Just not sure what the right config for F5 (version 10.4) should

Does `cf logs --recent` work for you? The recent logs request goes over HTTP. If that goes through that means only the websocket requests to loggregator servers are a problem. Rohit

Based on your description, it doesn't sound like warden networking or the warden iptables chains are your problem. Are you able to share all of your routes and chains via a gist? route -n ifconfig

Thanks Mike! Unfortunately, upgrading is not an option since its a really loaded enterprise device. The interesting part is that there is a similarly set up websockets vip (plain old server i think

-- Thanks & Best Regards, chunhua, zhang（张春华） M: ＋86 187 5198 6615 Department: CONSULTING Manager: Leon Cheng IT issue? Mail to: ask(a)pivotal.io

Great. Let us know if you have further questions, or share if you end up getting something cool deployed and working, since I haven't seen Spark on BOSH or CF yet. Best, Amit

I'm curious about the reasoning for the migration out of NATS, there was any limitation you guys hit?

We are running 11.4 and 11.6. I'd give an upgrade a try before digging too much deeper. Mike