831 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
Lumpiasty	f4f7bb8bab	Merge pull request 'Update Helm release open-webui to v14.6.0' (#293) from renovate/open-webui-14.x into fresh-start ... Reviewed-on: #293	2026-05-22 14:09:13 +00:00
Lumpiasty	ad80b5703f	pin version of kaneo image (default is latest)	2026-05-22 03:51:34 +02:00
Renovate	785bf6e8a4	Merge pull request 'Update dependency kaneo to v2.7.4' (#297) from renovate/kaneo-2.x into fresh-start	2026-05-22 01:39:14 +00:00
Renovate	12370b1886	Update dependency kaneo to v2.7.4	2026-05-22 01:39:11 +00:00
Lumpiasty	8f41e43340	Add Github token to renovate cron	2026-05-22 03:37:08 +02:00
Lumpiasty	c161da3657	add mlock and disable mmap in llama-server	2026-05-21 23:18:05 +02:00
Lumpiasty	fc2c15d154	move whisper to gpu	2026-05-21 22:02:34 +02:00
Lumpiasty	02b3ec13b4	switch kokoro to remsky/Kokoro-FastAPI	2026-05-21 21:55:34 +02:00
Lumpiasty	989732e1b5	move kokoro to separate deployment	2026-05-21 21:34:33 +02:00
Lumpiasty	ab438be629	fix tts model path	2026-05-21 21:11:29 +02:00
Lumpiasty	4556ca3c08	add ffmpeg for whisper	2026-05-21 20:58:56 +02:00
Lumpiasty	611f9f3886	add tts and sst to llama-swap and openwebui	2026-05-21 20:43:54 +02:00
Renovate	d1c689b149	Merge pull request 'Update ghcr.io/mostlygeek/llama-swap Docker tag to unified-vulkan-2026-05-21' (#296) from renovate/ghcr.io-mostlygeek-llama-swap-2026.x into fresh-start	2026-05-21 17:40:44 +00:00
Renovate	92bf792320	Update ghcr.io/mostlygeek/llama-swap Docker tag to unified-vulkan-2026-05-21	2026-05-21 17:40:41 +00:00
Lumpiasty	a2d27c5eee	fix llama-swap auto update	2026-05-21 19:40:03 +02:00
Renovate	75a183aeba	Update renovate/renovate Docker tag to v43.191.2	2026-05-21 17:27:26 +00:00
Renovate	0968f82345	Merge pull request 'Update Helm release openbao to v0.28.3' (#295) from renovate/openbao-0.x into fresh-start	2026-05-21 17:27:26 +00:00
Renovate	41b8cd3120	Update Helm release openbao to v0.28.3	2026-05-21 17:27:22 +00:00
Renovate	0ddf591828	Merge pull request 'Update quay.io/openbao/openbao Docker tag to v2.5.4' (#292) from renovate/quay.io-openbao-openbao-2.x into fresh-start	2026-05-21 02:05:27 +00:00
Renovate	4c2a3e2613	Update Helm release open-webui to v14.6.0	2026-05-21 02:05:26 +00:00
Renovate	7834cb4e43	Merge pull request 'Update Helm release woodpecker to v3.6.3' (#291) from renovate/woodpecker-3.x into fresh-start	2026-05-21 02:05:24 +00:00
Renovate	888c89586e	Update quay.io/openbao/openbao Docker tag to v2.5.4	2026-05-21 02:05:22 +00:00
Renovate	e94c7a7c78	Merge pull request 'Update ghcr.io/fluxcd/flux-cli Docker tag to v2.8.8' (#290) from renovate/ghcr.io-fluxcd-flux-cli-2.x into fresh-start	2026-05-21 02:05:19 +00:00
Renovate	fa0f0fc7c0	Update Helm release woodpecker to v3.6.3	2026-05-21 02:05:18 +00:00
Renovate	a519f4fe84	Update ghcr.io/fluxcd/flux-cli Docker tag to v2.8.8	2026-05-21 02:05:14 +00:00
Lumpiasty	4f74cdd092	Merge pull request 'Update Helm release openbao to v0.28.2' (#280) from renovate/openbao-0.x into fresh-start ... Reviewed-on: #280	2026-05-20 18:24:38 +00:00
Lumpiasty	313cd039e3	Merge pull request 'Update renovate/renovate Docker tag to v43.186.1' (#281) from renovate/renovate-renovate-43.x into fresh-start ... Reviewed-on: #281	2026-05-20 18:24:11 +00:00
Lumpiasty	cae91f992e	Merge pull request 'Update Helm release gitea to v12.6.0' (#284) from renovate/gitea-12.x into fresh-start ... Reviewed-on: #284	2026-05-20 18:24:02 +00:00
Lumpiasty	cfa3df6d1a	increase llama models PVC from 300Gi to 400Gi	2026-05-20 19:49:25 +02:00
Lumpiasty	c82f60e90a	switch text encoder to ponpoke/flux2-klein-4b-uncensored-text-encoder Q4_K_M	2026-05-20 17:50:06 +02:00
Lumpiasty	b41342be01	switch image model to FLUX.2-klein-4B (Apache 2.0, 4-step, unified gen+edit)	2026-05-20 16:49:30 +02:00
Renovate	55135cfaab	Update renovate/renovate Docker tag to v43.186.1	2026-05-20 02:00:46 +00:00
Lumpiasty	25a7b6c242	remove dns from mgmt interface on dlink	2026-05-20 01:43:17 +02:00
Lumpiasty	d3434a4102	remove unused qwen nothink chat template	2026-05-20 01:38:24 +02:00
Lumpiasty	4748ba3f73	enable image editing in openwebui via sd-server /v1/images/edits	2026-05-20 01:29:18 +02:00
Lumpiasty	de2822fee1	switch llama-swap to unified-vulkan image with FLUX.1-dev image generation ... - Update deployment to unified-vulkan-2026-05-19 (includes llama-server, sd-server, whisper-server in one image) - Fix binary paths: /app/llama-server -> llama-server (now on PATH) - Migrate groups -> matrix to allow FLUX to evict the always-on 0.8B model when image generation is requested - Add FLUX.1-dev Q4_K_S model via sd-server - Configure OpenWebUI image generation to use llama-swap sd-server - Update renovate versioning regex to treat all unified-vulkan date tags as patch updates for automerge	2026-05-20 01:11:57 +02:00
Lumpiasty	55ac337a63	enable MTP on MTP models	2026-05-18 20:31:57 +02:00
Lumpiasty	c3e5931c50	Merge pull request 'Update ghcr.io/mostlygeek/llama-swap Docker tag to v216' (#289) from renovate/ghcr.io-mostlygeek-llama-swap-216.x into fresh-start ... Reviewed-on: #289	2026-05-18 18:00:04 +00:00
Renovate	cd8768de67	Update ghcr.io/mostlygeek/llama-swap Docker tag to v216	2026-05-18 17:59:16 +00:00
Lumpiasty	5397749a73	Merge pull request 'Update ghcr.io/mostlygeek/llama-swap Docker tag to v214-vulkan-b9174' (#288) from renovate/ghcr.io-mostlygeek-llama-swap-214.x into fresh-start ... Reviewed-on: #288	2026-05-18 17:57:20 +00:00
Lumpiasty	f3ad488bc8	add MTP version of Qwen3.6-35B	2026-05-18 19:42:48 +02:00
Renovate	8f51671c35	Update ghcr.io/mostlygeek/llama-swap Docker tag to v214-vulkan-b9174	2026-05-17 12:46:14 +00:00
Renovate	0e5a830651	Merge pull request 'Update alpine/k8s Docker tag to v1.36.1' (#287) from renovate/alpine-k8s-1.x into fresh-start	2026-05-17 12:45:26 +00:00
Lumpiasty	a59c0e7c4f	Merge pull request 'Update ghcr.io/mostlygeek/llama-swap Docker tag to v214' (#286) from renovate/ghcr.io-mostlygeek-llama-swap-214.x into fresh-start ... Reviewed-on: #286	2026-05-17 12:45:22 +00:00
Renovate	a8b93399ff	Update alpine/k8s Docker tag to v1.36.1	2026-05-17 12:45:22 +00:00
Renovate	38ad78e69b	Update ghcr.io/mostlygeek/llama-swap Docker tag to v214	2026-05-17 02:01:24 +00:00
Lumpiasty	2d69cc6569	fix(ansible): resolve LTE failover data-plane bug on BroadMobi BM806C ... The embedded BroadMobi BM806C modem (Qualcomm MDM9225, firmware M1.2.0_E1.0.1_A1.1.8) in the D-Link DWR-921 C1 has two independent firmware bugs that together break the QMI data plane: 1. Modem accepts 802.3 framing but its 802.3 path is buggy — downlink frames never reach the host kernel. raw-ip framing works. 2. qmish calls uqmi --start-network --apn <foo>, which triggers FS#1363: the modem establishes a phantom bearer that gets IP addresses but has no working data path. Using --start-network --profile <N> (referencing a pre-configured NVRAM profile with the same APN) works correctly. Fixes applied: - qmish patches (3x community.openwrt.lineinfile): * Replace --set-data-format 802.3 with raw-ip * Replace --wda-set-data-format 802.3 with raw-ip * Bracket raw_ip sysfs write with ip link down/up (kernel rejects write with -EBUSY when wwan0 is already up) - Modem NVRAM: create/modify profile 2 (internetipv6, ipv6) for the IPv6 APN — profile 1 is already managed by qmish's --modify-profile - UCI wwan: add profile=1 and v6profile=2 so qmish uses --start-network --profile instead of --apn on both the v4 and v6 legs - Firewall: add wwan zone (input REJECT, output ACCEPT, forward REJECT) and Allow-ICMPv6-wwan rule - main.yml: reorder — packages (including usb-modeswitch) now run before wwan setup, so the modem is out of EDL mode when wwan.yml queries it for profile creation See docs/wwan-bm806c-qmi-workaround.md for the full diagnosis (what we ruled out, how we confirmed, manual setup steps, component versions, future upstreaming).	2026-05-16 21:20:26 +02:00
Lumpiasty	92447996fc	feat(ansible): migrate OpenWrt package management from opkg to apk ... OpenWrt 25.12+ uses apk instead of opkg. The community.openwrt.apk module is only available in the unreleased 1.4.0 (git main), so requirements.yml now installs the collection from git. - requirements.yml: install community.openwrt from git main branch (comment explains why — apk module not yet in a Galaxy release) - packages.yml: switch from community.openwrt.opkg to apk, use join filter to pass all packages at once instead of looping	2026-05-16 21:20:24 +02:00
Lumpiasty	dd559ade43	feat(ansible): Enable WLAN LED on dlink	2026-05-16 21:20:20 +02:00
Renovate	8ce9462d4a	Update Helm release gitea to v12.6.0	2026-05-16 02:01:22 +00:00