Understand queue hotspots

Identify scheduler queues whose waiting work is becoming unhealthy.

Last checked 21 August 2026

The Hotspots card turns several Queue Status measurements into a shorter investigation list. It includes queue-wide backlogs and specific websites or pages that cross fixed load thresholds.

Hotspots are operational signals. They do not create Advisory, Warning, or Critical incidents and do not change Health Score.

Queue backlog

A queue receives a Hotspot badge and a backlog row when:

waiting jobs >= max(8, workers × 3)

This prevents a tiny burst from being labelled a backlog while scaling the threshold for queues with more workers. Examples:

one worker: hotspot at eight waiting jobs
two workers: hotspot at eight waiting jobs
four workers: hotspot at 12 waiting jobs
six workers: hotspot at 18 waiting jobs

The hotspot detail shows the queue label and current number of jobs waiting.

Identify the lane before acting. A Page checks backlog points toward page count, schedules, slow destinations, or crawl concurrency. Search Console lanes can be constrained by their own quotas and rate controls. SSL, Domain, DNS, and Site resources use separate schedulers. Screenshots and Icons can be deferred while page checks are backlogged so monitoring crawls remain the priority.

Slow page check

A page appears as Slow page check when its latest known check duration is at least 15 seconds. TechDash uses the latest stored check-run duration, falling back to the queue's most recent duration when needed.

The row shows the page path and duration. The card lists at most five slow pages, ranked from longest to shortest. Slow output can come from the destination server, network, redirects, browser rendering, challenged access, or a resource-heavy page. Check the page directly and review its recent run evidence before changing concurrency.

Large HTML capture

A page appears as Large HTML capture when its stored compressed HTML is at least 250,000 bytes. The displayed value is gzipped size, not the uncompressed response body or the page's complete transferred weight.

The card lists at most five large pages, ranked by compressed stored bytes. Large captures increase local storage and processing pressure. Investigate unexpectedly large templates, embedded data, duplicated markup, and pages returning a different document to the crawler.

Heavy check schedule

A website appears as Heavy check schedule when its enabled pages are estimated to create at least 200 checks per day.

Each enabled page contributes this daily rate:

Every 15 minutes: 96 checks per day per page
Hourly: 24
Daily: 1
Weekly: approximately one seventh, or 0.14, per day

TechDash sums these daily page rates, including fractional weekly rates, then rounds the site total once. For a site with only weekly pages:

One weekly page contributes approximately 0.14 checks per day, so the displayed site total is 0.
Four weekly pages contribute approximately 0.57 checks per day, so the displayed site total is 1.
Seven weekly pages contribute 1 check per day, so the displayed site total is 1.

The hotspot shows enabled pages and the rounded estimated checks per day. A mixed schedule is calculated from the actual interval on every enabled page before the site total rounds.

Many monitored pages

A website appears as Many monitored pages at 40 or more enabled pages, but only when it is not already shown as Heavy check schedule. This avoids two broad schedule hotspots for the same site.

Forty pages is a load threshold, not a product limit. More pages can be appropriate when their check frequencies and computer capacity are sensible.

Decide whether to act

Watch the Queues table over multiple 10-second refreshes. A temporary hotspot after startup, wake, reconnection, bulk add, or manual runs can clear naturally. Falling Waiting Jobs means the lane is recovering.

Investigate when waiting is sustained or growing, pages repeatedly exceed thresholds, or last-check times are becoming stale. Start with the linked website, then reduce low-value monitored pages or frequency. Change Concurrent Crawls only for page-check capacity, and only after considering CPU, memory, network, and destination limits.