LTS16 OPCH Release - Environment Variable Changes

Announcements
,
1

The LTS16 OPCH is released! With this release, we’ve standardized some of our environment variables, while also adding two others that customers can opt in to support troubleshooting of the OPCH. The grace period for customers on weekly releases starts with r377 and ends at r382. LTS customers are granted one LTS version grace period to upgrade all OPCH per usual.

You can read more about the upgrades below:

2

Thanks @canaalpaslan
I start checking my Docker configuration, I have note that with LTS16 I need to do the following change:
CONNECTORS_HTTP_PROXY => TULIP_CONNECTORS_HTTP_PROXY
CONNECTORS_HTTPS_PROXY => TULIP_CONNECTORS_HTTPS_PROXY

What about those I have in env? It look like I don’t need to change:
NO_PROXY
NODE_TLS_REJECT_UNAUTHORIZED

Last, I very curious of the new setting you have introduce:
-e TULIP_LOGGING_TO_TULIP=true
-e TULIP_TELEMETRY_TO_TULIP=true
What kind of log or troubleshooting help it give?

Thanks a lot

3

Hi @Nicolas,

Correct — NO_PROXY and NODE_TLS_REJECT_UNAUTHORIZED don’t need to change.

On TULIP_LOGGING_TO_TULIP and TULIP_TELEMETRY_TO_TULIP: when enabled, the OPCH sends its logs and diagnostic metrics directly to Tulip. The logs are the same ones you see when you open an OPCH’s logs; the metrics cover CPU, memory usage, and network throughput (e.g. 1.5 MB/s in/out). Together they let us spot a struggling OPCH — for example, one maxing out memory usually shows other side effects too.

These variables help troubleshooting in two ways:

  1. When you report an OPCH problem, we already have the logs so there is no need to send them over. Investigations often fall into a loop of change something, test, request logs, change again, request logs again, which is slow and frustrating for everyone. With logs already in Tulip Cloud, that goes away.
  2. Once the logs and metrics enter our standard aggregation system, we can run automatic checks for irregularities and alert our Support/Engineering team early. For Cloud services already using this system, we often catch and fix issues before a customer is negatively impacted.

Best,
Cana