🧵 The second issue is with the

CleanupExpiredHosts

cron job (the

expired_hosts

job within

cleanups_then_aggregation

). Our setup: • ~9M hosts in Fleet, only ~100k actively checking in • Host expiry is enabled, but stale hosts aren't being removed Error:

get host details for expired hosts: select hosts by id: Error 1390 (HY000): Prepared statement contains too many placeholders

Root cause: Three code paths pass the full list of expired host IDs into unbounded

IN (?)

clauses, which blows past MySQL's ~65,535 placeholder limit: •

ListHostsLiteByIDs

•

DELETE FROM host_seen_times

•

CleanupHostActivities

Related issues: • This looks related to github.com/fleetdm/fleet/issues/31157 which was reported by another customer and closed. The symptoms are the same — hosts not being removed after hitting the expiry window — but at our scale the prepared statement limit is the specific failure mode. • Also wanted to ask about github.com/fleetdm/fleet/issues/45947 — is there a timeline or release planned for a fix? Our fix: I've wrapped the three unbounded

IN

clauses with

BatchProcessSimple

. github.com/adityaoza1901/…/cleanup-expired-hosts-batching

Hello, github.com/fleetdm/fleet/issues/45947 should be part of 4.89.0. You can see this on the right-hand side of the page for the issue.

If you have, please open a pull request detailing what you ran into and how you resolved it. Our team would love to review it and merge it if it fixes a bug.

@Steven Palmesano I was on a vacation. Just got back. Let me get back to you on this later this week. For now, I have another issue I'd like to report. Will do so in a new thread.

Thanks. I'll open the bug report for this today. When you submit the PR, the template will have a spot for you to enter the issue number that it resolves, which will connect it to the bug report.

Here it is! github.com/fleetdm/fleet/issues/49730