Troubleshooting: Overview

In SPG99, almost any problem is easiest to investigate in the same order. This saves time and quickly separates an application error from a platform issue.

Step 1. Identify the symptom

Usually, the problem belongs to one of these groups:

• the database is taking too long to create or start;
• connection fails because of TLS / DSN;
• cold start feels too slow;
• the error is related to the token or permissions;
• the database moved into error.

Step 2. Look at the resource state

The first things to check are:

• the database state;
• the tenant and database name;
• whether there has been recent activity;
• whether there is an obvious deleting or error.

Step 3. Check the entry layer

If the problem looks like “I can’t connect,” make sure that:

• you are using the Gateway host from the DSN;
• the DSN does not contain an API key;
• TLS is enabled;
• connect_timeout is not overly aggressive.

Step 4. Open Monitoring

Look at:

• Metrics — to understand whether this is cold start, load, or a system problem;
• Logs — to see the actual error;
• Active queries — if the problem looks like “the database is stuck” or “everything is slow.”

Step 5. Prepare context for support

A good set of data includes:

• tenant and database name;
• time of failure;
• the error text;
• a log fragment;
• what exactly you were doing before the problem.

Common scenarios

Below in this section there are separate breakdowns for the most frequent cases:

• Database stuck in creating / booting;
• Cannot connect / TLS;
• Cold start is too slow;
• Common errors reference.