rinkhals-community/Rinkhals 20261022_01

Hello my wonderful Rinkhalsnauts!

This is a test release for October. A lot of community work landed in this one, so please give it a spin and tell us what you find... especially if you own a Kobra S1 Max or a Kobra 2 Pro, or print through SimplyPrint.

Highlights

Prints started from the slicer or the cloud no longer fail about 30 seconds in. Rinkhals' metadata step could rewrite a G-code file while GoKlipper was already reading it, so the printer resumed mid-line and stopped with an "abnormal end of code" style error. The file is now replaced atomically, so a print that has already started keeps reading the original. This was a long-standing race, not something new in 20260901_01. If you added [!mmu_ace] to moonraker.custom.conf as a workaround, you should be able to remove it after updating.
Per-gate Spoolman spools for the ACE. Each ACE gate can now be bound to a Spoolman spool, and the binding survives status updates, Moonraker restarts and reboots. The active spool follows whichever gate is loaded at print start and at every tool change.
Finished prints are recorded as completed, not cancelled. GoKlipper reports "standby" the moment the file is consumed and only says "complete" after its end-of-print routine, so Mainsail and Fluidd were logging every finished print as cancelled. Rinkhals now keeps reporting the print as running until the printer really is done.
Bed mesh calibration from Mainsail fits each printer. The calibration script is now built from the commands your printer actually has. On the S1 family it heats over the purge bin, wipes, and then stays parked on the silicone pad while it cools, instead of oozing at the wiper exit.
The Heightmap shows the mesh you are actually using, including adaptive meshes, on the Kobra S1 with firmware 2.7.2.7.
Kobra 2 Pro 3.1.4: the touchscreen no longer dies when you leave the Rinkhals menu. The K3SysUi patch passed the wrong object pointer, which crashed the Anycubic UI when you returned to it.

Behavior changes to be aware of

If you have [spoolman] configured, the spool picker now appears in Mainsail and Fluidd. There is nothing to switch on.
The spool ID Rinkhals used to make up from the RFID SKU is gone. Spools are now real Spoolman spools you assign.
Spools are not forgotten automatically. spool_forget_after is off by default. To have a gate drop its spool after it has been empty for a while, add for example spool_forget_after: 300 to [mmu_ace]. Putting a different RFID product into a gate still clears its assignment. spool_auto_match (off, unique or latest) is also available and off by default.
A print that uses a tool mapped to an empty ACE gate is refused before it starts, with a message telling you which tool and gate, instead of failing hours later with "Filament is runout". Some prints that used to start (and then fail) will now be refused up front.
"Print finished" notifications arrive when the printer actually stops moving, about 25 seconds later than before on the S1.

Printing and calibration

Large OrcaSlicer files no longer time out during metadata processing. The limit went from 20 seconds to 300.
MMU_TTG_MAP accepts a partial map such as MAP=0,2 instead of crashing or silently resetting, and when used mid-print it tells you it applies from the next print.
The bed mesh profile load arrow in Mainsail works on GoKlipper, and saving a profile named default is no longer refused.
The K3 re-homes Z with the nozzle hot before probing, for a more accurate mesh.
New console command MMU_GATE_SKUS.

Mainsail, Fluidd and Moonraker

G-code console output such as PROBE_ACCURACY now shows up in Mainsail and Fluidd.
Saving a config file from Fluidd can no longer truncate it to zero bytes.
Fluidd no longer freezes on ACE status updates, and filaments are editable in Fluidd again.
Object queries return only the objects asked for. This fixes the BIQU Panda Breath "Auto" mode, and clients that poll no longer swallow ACE updates meant for others.
The RFID temperature messages that filled most of the Moonraker log are now only logged when they change.
Stopping Moonraker no longer leaves its supervisor running, which could end up with two Moonrakers.
Moonraker now gets its full ten seconds to shut down cleanly.

Installer and web portal

"Manage system versions" in the installer works again. The old firmware mirror was gone; it now reads the Rinkhals.Firmwares manifests.
A failed Rinkhals update is reported as a failure, both in the web portal and on the touchscreen, instead of looking like it succeeded.
Vanilla ACE no longer restarts Moonraker on every start on the Kobra 2 Pro 3.1.4.

Please help us test

Kobra S1 Max owners: please run a bed mesh calibration from Mainsail (Heightmap, Calibrate) and tell us whether the nozzle stays parked on the pad while it cools. This was verified on an S1 but not an S1 Max.
SimplyPrint users: please check that a finished print shows as completed, and that the touchscreen leaves the printing screen.

Thank you

This release is mostly community work. Thank you to rd65, kpaaro, jschroeck, metroiss, jansobczak, 7487, Felixoid, and to everyone whose reports and debug bundles made the fixes possible: INDIAJULIETVICTOR, russdan and smeroni68 (#128), 1walkerit (#126), Hello-dot-Jpg (#140), ChuChuCharlie (#178), M4kI9H7 (#101), g-h-u-l (#131) and mane (#170).

Known issues

These are open and not fixed in this release.

#111 - a rare stall in GoKlipper's status query path can shut down the MCU mid-print on the Kobra 3 Max.
#143 - OctoEverywhere stops connecting when LAN-only mode is turned on (KS1M).
#134 - a report of an ACE tool change being silently dropped mid-print. We are waiting on a debug bundle.

Report findings to https://github.com/rinkhals-community/Rinkhals/issues with a debug bundle.

Full Changelog: 20260901_01...20261022_01_test

rinkhals-community/Rinkhals 20261022_01_test Rinkhals 20261022_01_test on GitHub