All SEs in the GCP user group
gcp-customer-support-engineering have access to managed instance via GCP management tooling. To request access to this group, reach out to #it-tech-ops to request access to the
A quick note before reading the rest of this page: Customers with managed instances cannot pull their own logs, and in many cases, the customer we’re dealing with is not a site-admin, so they can’t access site-admin privileges either. Please make absolute sure regardless of the customer, that they are not working with a managed instance.
A list of managed instances can be found as part of the GCP Project
If you are an support engineer being trained to work with managed instances, read ahead.
If you do not have access, create a ticket in #it-tech-ops request to be added to
gcp-customer-support-engineering Google Group and tag your manager for approval.
After you’ve followed the docs in the ‘required access’ section and received the access to all necessary credentials, reach out to Jason or Mariam to set up a thirty minute tutorial that will take you through the most important aspects of accessing managed instances.
If all of us happen to be outside of working hours when you need a tutorial, watch this video as a supplement. That said, it would still be good to schedule some time to meet with one of us when we’re back online.
Next, you’ll need to install the Google Cloud Platform (GCP) SDK locally. Instructions for doing so can be found here. Pay close attention to the troubleshooting tips in this link, as you are likely to run into a few minor snags along the way.
To check and see if you’ve installed the GCP SDK correctly, enter the command
gcloud version in your terminal. It should return the current version of your SDK. If so, congratulations. You now have the ability to start up virtual machines for each managed instance, and begin debugging.
To access managed instance’s running external databases (Cloud SQL) via
psql you’ll need to install
cloud_sql_proxy, instructions to do so may be found here.
Next, become familiar with the managed instances operations page. This provides overview for debugging managed instances, and is absolutely necessary reading. Bookmark the page. You will revisit it often.
In general, this is not advised. While you will have access to every instance managed by Sourcegraph, we want to limit this approach in order to maintain the privacy of our customers. However, if other approaches fail to resolve an issue, this is an option.
Should you find yourself in a situation where you need to login to a customer’s instance directly, make sure you inform the customer before you do so, and provide thorough and clear context for your reasoning behind needing to do so. Be completely transparent about your intentions so that you and the customer are on the same page.
If while working on an issue for a managed instance customer you encounter and issue that customer support cannot resolve, engage the cloud team in the
#cloud slack channel. After gathering relevant data, and checking with team CS team members for ideas.