Interpretation
How do you interpret the customer's words?
The customer uses pre-install queries to scope software installs to a subset of hosts (for example, by timezone). When the query returns no results, Fleet correctly skips the install script, but the attempt is still logged as a failed install. This floods the host activity feed with repeated “failed to install” entries, inflates the software title's failure count, and causes anything keyed on failures (retries, alerting, reporting) to treat an expected, by-design outcome as an error.
What's Fleet missing?
Fleet doesn't distinguish between an install that actually failed to run and an install that was correctly never attempted because its pre-install query condition wasn't met. Both currently produce the same “failed” activity and status.
What does the customer's ideal workflow look like?
When a pre-install query returns no results, Fleet either records no activity at all (the customer's preferred outcome — silent until the host actually qualifies and installs successfully), or logs a distinct, non-failure status such as “Skipped” / “Condition not met.” Either way, this status is excluded from the software title's failure count and from anything that alerts or retries on failures, so admins scanning for real install problems only see genuine errors.
Interpretation
How do you interpret the customer's words?
The customer uses pre-install queries to scope software installs to a subset of hosts (for example, by timezone). When the query returns no results, Fleet correctly skips the install script, but the attempt is still logged as a failed install. This floods the host activity feed with repeated “failed to install” entries, inflates the software title's failure count, and causes anything keyed on failures (retries, alerting, reporting) to treat an expected, by-design outcome as an error.
What's Fleet missing?
Fleet doesn't distinguish between an install that actually failed to run and an install that was correctly never attempted because its pre-install query condition wasn't met. Both currently produce the same “failed” activity and status.
What does the customer's ideal workflow look like?
When a pre-install query returns no results, Fleet either records no activity at all (the customer's preferred outcome — silent until the host actually qualifies and installs successfully), or logs a distinct, non-failure status such as “Skipped” / “Condition not met.” Either way, this status is excluded from the software title's failure count and from anything that alerts or retries on failures, so admins scanning for real install problems only see genuine errors.