3D print farm maintenance checklist (by cadence)

A 3D print farm maintenance checklist has four cadences: daily checks (nozzle, bed adhesion, filament spool status), weekly tasks (belt tension, lubrication, fan dust-out), runtime-triggered tasks fired by print hours, filament weight, print count, or failure rate, and on-demand tasks when a printer flags a problem. The runtime-triggered tier is the one that separates a print farm maintenance routine from a hobbyist one: on a fleet of 20 Bambu A1s running unevenly across orders, a calendar-only schedule misses the machines doing 18 hours of prints a day and over-services the ones that sit idle for a week.

This 3d print farm maintenance checklist walks through all four cadences with concrete checks for each, a spare-parts inventory you should keep on a shelf, and the operational rule that matters most: a printer in maintenance has to leave the queue while it is being serviced, automatically, or your queue keeps sending it work. By the end you will have a complete 3d printer maintenance routine that scales from ten printers to several hundred without becoming a spreadsheet job.

What does a 3D print farm maintenance checklist actually look like?#

A print farm maintenance routine is four nested layers, each with its own trigger:

1. Daily - a morning glance that takes 30 to 60 seconds per printer. Visual checks, no tools.
2. Weekly - slightly deeper tasks scheduled per printer group, usually 5 to 10 minutes per machine.
3. Runtime-triggered - tasks that fire on print hours, filament weight, print count, or failure rate, scoped per printer or per model.
4. On-demand - flagged problems (printer reports a fault, AI detection escalates, an operator marks a failure) that create a maintenance job immediately.

The first two are easy: pick a checklist, do the checklist. The third is where most farms have a gap, because a calendar reminder ("change the nozzle every month") is wildly wrong for a fleet with uneven utilisation. The fourth needs a mechanism to take a printer out of rotation while you fix it, which is the operational rule that ties the whole system together.

Daily checklist (morning glance)#

These run on every printer, every morning, before the day's queue gets dispatched. They are visual and take seconds.

• Nozzle tip: any oozed filament, stringing residue, or visible blob. Pick it off before the next print starts. A clean nozzle is a clean first layer.
• Bed surface: scraps from a failed clear, leftover prime line, or print residue. Wipe with IPA if the print surface is PEI; replace the plate if it is PEI sheet and worn.
• Filament spool: enough filament left for the next planned print, no tangles on the spool, no obvious moisture-related foam. If you run an AMS or material station, glance at each slot.
• Cooling fans: spin freely, no audible bearing complaint. A clicking part-cooling fan today is a failed print tonight.
• Belts and frame: no obvious slack, no Z-rod backlash, no loose bed clips. Don't tighten anything yet, just note it for the weekly task.
• The dashboard: any printer flagged as overdue, in maintenance, or out of order. That is your prioritised work list for the day.

A morning glance is not "everything is fine, time to print". It is "what is the worst thing about the farm right now and is it bad enough to handle before the queue starts dispatching?" Most days the answer is no, and the daily check takes ten minutes for a 20-printer farm. The days the answer is yes are the days you would have lost six hours of print time to a preventable failure.

Weekly checklist (per printer group)#

Weekly tasks sit on a calendar. They are the textbook 3d printer maintenance schedule items that vendor manuals cover, but on a farm you want them grouped by printer family so the schedule applies to the right hardware.

• Belt tension: check X and Y belts (and Z for belt-driven Z, e.g. CoreXY). Tension to spec; loose belts cause layer shifts, over-tight belts wear pulleys.
• Linear rails / smooth rods: wipe and re-grease per the manufacturer's interval. PTFE-friendly grease for hardened steel rods, lithium for V-slot wheels.
• Lead screws: clean and re-lube (a single drop of light machine oil is usually enough).
• Fan grilles, electronics box, PSU vents: blow out with compressed air. Dust in the PSU is the cause of "the farm randomly resets at 3am" issues.
• Hotend socks: replace if torn or saturated. A clean sock holds the temperature stable; a slumped sock causes heat creep and clogs.
• Build plate inspection: PEI, glass, or textured plate? Each wears differently. Note plates that need replacing soon and add them to your spare-parts queue.
• Cable inspection on moving parts: look for kinks, fraying, or insulation wear on the cable carrier (the "umbilical" between the toolhead and the printer body). Replace before the wire breaks.
• Firmware: are you on the version you want to be on? If you do firmware updates as a fleet, do them on the weekly window, not mid-day.

Group these per printer model. Bambu A1, Prusa MK4S, Creality K2 Plus all have slightly different lubrication points and belt tension procedures. A schedule scoped to model: Bambu A1 runs the right tasks on the right machines automatically; a company-wide schedule does not.

Runtime-triggered maintenance (the real differentiator)#

This is where a print farm maintenance routine diverges from a hobbyist one. Calendar reminders work fine when you have one printer running a couple of hours a day. They break completely on a fleet where some printers run 18 hours a day and some run two.

The right model is to fire maintenance off the actual usage of each individual printer. SimplyPrint schedules support six trigger types out of the box:

• Calendar (days). Every N days since last completion, per printer. The classic calendar trigger, useful for weekly and monthly checks.
• Print hours. Every N hours of print time on that specific printer. The prediction engine learns each printer's average daily usage and creates the job ahead of when it will be due, so your team has lead time.
• Filament weight. Every N grams of filament pushed through. The single best proxy for hotend wear, especially with abrasive filaments.
• Print count. Every N completed prints. Useful for tasks that scale per-job rather than per-hour (cleaning the bed of leftover skirt residue, for example).
• Failure count. After N print failures on a printer, with optional filtering by cancel reason (so user-cancels don't trigger maintenance, but layer shifts and clogged-nozzle failures do).
• Bundled tasks. Combine several due tasks into one efficient job, so technicians don't get pulled into the same printer three times in a week.

An example schedule for a fleet of 20 Bambu A1s#

To make this concrete, here is a runtime-triggered maintenance schedule for a 20-printer Bambu A1 farm running mixed PLA and PETG, with occasional abrasive runs.

• Daily glance: time-based, every 1 day, scope: company. Just the morning checklist above as a single checklist task.
• Weekly inspection: time-based, every 7 days, scope: model Bambu A1. Belt check, fan dust-out, hotend sock inspection, plate condition.
• Nozzle inspection: filament-grams trigger, every 500 g, scope: model Bambu A1. Catches abrasive-filament wear. For non-abrasive runs the cumulative grams stack up slower, so the schedule self-paces.
• Hotend rebuild: print-hours trigger, every 800 print hours, scope: model Bambu A1. This is a sensible default 3d printer nozzle replacement interval for brass nozzles on non-abrasive filament; the prediction engine flags the job 7 days in advance so spare parts can be ordered if stock is low. Bambu's own hotend assembly maintenance guide is a useful supplement for the actual swap procedure on Bambu hardware.
• Plate replacement check: print-count trigger, every 200 prints, scope: model Bambu A1. PEI plates wear out around 300 to 500 prints; this fires a "look at it, replace if needed" task before the next print fails.
• Struggling-printer audit: failure-count trigger, after 5 failures (filtered to mechanical reasons), scope: company. Any printer that fails 5 times for a real reason in a short window gets pulled out of rotation for inspection automatically.

Six schedules, the entire fleet covered. Note that none of them touch the Bambu AMS interval, because that is vendor-specific and lives in Bambu's documentation, not the schedule engine.

The combination of model-scoped schedules and per-printer counters means a farm with mixed hardware does not need separate spreadsheets per printer family. The same dashboard shows next-due jobs for every machine, predicted from each machine's actual usage.

On-demand maintenance (when a printer flags a problem)#

The first three cadences are scheduled. The fourth is reactive, and on a busy farm it is what eats your day if you let it.

Three things create an on-demand maintenance job:

• A printer reports a fault: thermal runaway, heater error, sensor fault. The printer's firmware throws an error, the panel marks the printer as out of order, and a maintenance job can be linked from the error.
• AI failure detection escalates: a print is auto-paused because spaghetti, layer shift, or adhesion failure was detected. If you have that hooked into AI failure detection, the operator confirms or dismisses; a confirmed failure can be tied to a maintenance job for the printer.
• An operator reports a problem: a problem-report flow exists for anyone with permission. They click "report a problem" on the printer, pick a category (mechanical, electrical, calibration, firmware, other), and a maintenance job is created. Useful when a student or junior operator sees something off but is not the person who fixes it.

Each of those creates a job linked to the printer, with the relevant tasks pre-populated based on the problem category. The job sits in the maintenance queue and gets assigned, started, and completed like any other job.

Spare parts inventory you should keep#

A 3d printer spare parts list for a fleet is not "one of each, just in case". It is calibrated to the fleet size and the failure rate of each part. The numbers below are starting points for a 20-printer FDM farm running mixed PLA, PETG, and occasional ABS; scale up or down with fleet size.

• Nozzles: a dozen 0.4 mm brass per printer model on the fleet, plus a smaller stock of 0.2 mm, 0.6 mm, and hardened-steel for abrasive jobs. Brass nozzles are cheap, do not skimp.
• Build plates: at least one spare per printer, ideally two. PEI textured wears the fastest, smooth PEI second, PC and glass plates last longest.
• PTFE tubes: a metre of bowden PTFE per Bowden printer, and a stock of pre-cut hotend PTFE for direct-drive printers that use one (some don't).
• Hotend socks: a pack per hotend type. They cost almost nothing and protect the heater block.
• Bed levelling springs / silicone bumpers: a set per printer if you have springs (older Creality, MK3S). MK4 and Bambu use rigid mounts, skip this for those.
• Fans: one spare hotend fan and one spare part-cooling fan per printer family. The part-cooling fan is the one that fails first; the hotend fan failure is the one that causes the worst damage (heat creep, clog, sometimes a melted heatbreak).
• Hot-end socks and silicone heater block insulators: a pack per hotend family.
• Belts: a metre of GT2 belt per printer model, plus a set of pulleys.
• Drive gears: one or two extruder drive gears per extruder type.
• Thermistors and heaters: one of each per printer family, in case of an electrical failure.
• Cable carriers / drag chain links: a few links per cable carrier type, for the inevitable broken-off link.

In the maintenance feature, each part has a stock level, a low-stock threshold, and is deducted automatically when a maintenance task that uses it gets ticked off. Low-stock fires a notification and a webhook. That means you do not need a separate spreadsheet to know when to reorder: the inventory page tells you what is below threshold and the notifications feature wakes you up if a critical part runs out.

Maintenance mode and what it does to AutoPrint#

This is the operational rule that ties the whole maintenance system to the queue: a printer that is being serviced has to be invisible to the queue, automatically.

When a maintenance job starts and the "puts printer in maintenance" toggle is on, the printer enters maintenance mode the moment the job starts. While in maintenance mode:

• AutoPrint skips it. The dispatch check fails for that printer, so AutoPrint will not send the next queued job to it, even if it shows as operational.
• 1-Click Print skips it. A user clicking "send to this printer" gets blocked, not dispatched and queued.
• Queue matching will not match items to it. The tag-and-metadata matching engine treats a printer in maintenance as ineligible, so an unattended overnight queue cannot accidentally pick up a printer mid-nozzle-swap.
• The printer status reports IN_MAINTENANCE. Operators see it clearly on the dashboard; the printer status table groups it under the "in maintenance" bucket.

When the job completes (or is cancelled), the printer rejoins rotation automatically, unless another active maintenance job is still holding it in maintenance. There is no manual "put printer back in service" step.

This matters because the alternative ("just disable the printer manually") is fragile. Manual disables get forgotten when the technician finishes the job at 2am, the printer sits offline until someone notices, and you lose dispatch time. Tying the maintenance job to the maintenance mode means the rotation gate closes when work starts and opens when work ends, no human in the loop.

The same maintenance-mode signal feeds into the statistics page as downtime, so a printer that goes into maintenance more often than the rest of the fleet shows up as a downtime outlier. Useful for spotting the printer in the corner that is quietly eating more technician hours than its peers.

What this checklist does not cover#

This checklist is intentionally vendor-agnostic. A few things are vendor-specific and you should defer to the manufacturer's documentation rather than try to schedule them generically:

• Bambu AMS lubrication and cleaning intervals. AMS units have their own service cadence (PTFE replacement, gear cleaning, hub calibration) that Bambu publishes per model. Schedule those separately from your fleet-wide tasks.
• Prusa MMU3 calibration and maintenance. The MMU3 has its own filament-loading and selector calibration routine; Prusa's docs are the source of truth.
• Creality Sonic Pad / Creality Cloud-specific firmware updates. If you run Creality printers under Sonic Pad or Klipper, firmware management lives outside the printer's own UI.
• Vendor-specific firmware revision intervals. These come out unevenly; subscribe to the manufacturer's release notes and update on a scheduled window, not ad hoc.
• Anything covered by warranty service. If a printer is still under warranty and the failure is on the manufacturer to fix, get them to fix it. Track the RMA in maintenance jobs but don't do their work for them.

The general rule: this checklist covers the print-farm operational layer (queue, dispatch, fleet-level scheduling, spare-parts inventory). Hardware-specific maintenance routines stay with the hardware vendor's documentation. The two systems talk to each other through the print-hours and filament-grams counters, but you should not be replacing what the manufacturer says about their own machine.

For the broader context of running a farm unattended, the continuous-printing AutoPrint guide covers the queue, the four bed-clearing methods, and the failure-handling rules that maintenance mode plugs into. The /print-farms landing page covers the full operational stack: queue, AutoPrint, staggered start, maintenance, AI detection. If you have just stood up a farm and are looking for the next configuration step after AutoPrint, maintenance schedules are it.

A good benchmark: a well-configured 20-printer farm should not need more than one technician half-day per week to keep running, plus reactive on-demand jobs when something flags. If you are spending more than that, the runtime-triggered tier is probably not configured, and the schedules above are the place to start.