diff --git a/.claude/reference/proxmox-inventory.md b/.claude/reference/proxmox-inventory.md
index 2b9ab3fe..5e308aee 100644
--- a/.claude/reference/proxmox-inventory.md
+++ b/.claude/reference/proxmox-inventory.md
@@ -92,7 +92,7 @@ Channel 3:  A4 [32G] ──── A8 [32G]  ──── A12[ 8G ]     = 72 GB
 | VMID | Name | Status | CPUs | RAM | Network | Disk | Notes |
 |------|------|--------|------|-----|---------|------|-------|
 | 101 | pfsense | running | 8 | 4GB | vmbr0, vmbr1:vlan10, vmbr1:vlan20 | 32G | Gateway/firewall |
-| 102 | devvm | running | 16 | 24GB | vmbr1:vlan10 | 100G | Development VM + t3code Workstation host. 14G swap (8G /swapfile + 6G /swapfile2, grown 2026-06-10; swappiness=10). Capacity budget: ~4-5G RAM/active user, max ~3-4 concurrent active Claude sessions. NOT Terraform-managed. |
+| 102 | devvm | running | 16 | 24GB | vmbr1:vlan10 | 100G | Development VM + t3code Workstation host. 14G swap (8G /swapfile + 6G /swapfile2, grown 2026-06-10; swappiness=10). Capacity budget: ~4-5G RAM/active user, max ~3-4 concurrent active Claude sessions. NOT Terraform-managed. Disk controller: `virtio-scsi-single` + `scsi0 iothread=1,aio=threads` staged 2026-06-11 after the QEMU I/O stall (was `scsihw: lsi`, the only VM on the legacy path — see `docs/post-mortems/2026-06-11-devvm-qemu-io-stall.md`); applies at next cold stop→start. |
 | 103 | home-assistant | running | 8 | 8GB | vmbr0 | 64G | HA Sofia, net0(vlan10) disabled, SSH: vbarzin@192.168.1.8 |
 | 105 | pbs | stopped | 16 | 8GB | vmbr1:vlan10 | 32G | Proxmox Backup (unused) |
 | 200 | k8s-master | running | 8 | 32GB | vmbr1:vlan20 | 64G | Control plane (10.0.20.100) |
diff --git a/docs/architecture/compute.md b/docs/architecture/compute.md
index 0b808b55..0f6ff092 100644
--- a/docs/architecture/compute.md
+++ b/docs/architecture/compute.md
@@ -101,7 +101,12 @@ graph TB
 > PVE host (sources in `infra/scripts/`, install pattern per
 > `architecture/backup-dr.md`). Timer fires `OnBootSec=5min` +
 > `OnCalendar=hourly`, so any drift (config restore, manual `qm
-> set`, fresh clone) self-heals within the hour. Current caps:
+> set`, fresh clone) self-heals within the hour. The script compares
+> *normalized option sets*, so an unchanged config is a true no-op —
+> until 2026-06-11 a raw string compare (defeated by `qm config`'s
+> canonical key order) re-issued `qm set` hourly against running VMs,
+> live-rewriting QEMU throttle state via QMP (implicated in the devvm
+> I/O stall; see `post-mortems/2026-06-11-devvm-qemu-io-stall.md`). Current caps:
 > 102 devvm 60/60, 103 home-assistant 40/40, 200 k8s-master 100/60,
 > 201 k8s-node1 150/120, 202 k8s-node2 150/120, 203 k8s-node3 150/120,
 > 204 k8s-node4 150/120, 220 docker-registry 40/40.
diff --git a/docs/post-mortems/2026-06-11-devvm-qemu-io-stall.md b/docs/post-mortems/2026-06-11-devvm-qemu-io-stall.md
new file mode 100644
index 00000000..d62ed5ff
--- /dev/null
+++ b/docs/post-mortems/2026-06-11-devvm-qemu-io-stall.md
@@ -0,0 +1,116 @@
+# 2026-06-11 — devvm dead ~90 min: QEMU-internal I/O stall on the legacy LSI disk path
+
+## Impact
+
+- devvm (VM 102, the shared multi-user Claude Code workstation) effectively
+  dead 15:21–16:48 UTC (18:21–19:48 EEST): all ssh/tmux and t3 sessions for
+  wizard/emo/anca lost, every in-flight agent killed.
+- Detection was human (~90 min) — no `up{instance="devvm"} == 0` alert
+  exists (follow-up below).
+- Recovery was manual: kill of the wedged QEMU process + `qm start` (the
+  kill left no autopsy — see "What we could not prove").
+
+## Timeline (UTC; host journal runs EEST = UTC+3)
+
+- **15:01** — hourly `apply-mbps-caps` run live-rewrites VM 102's scsi0
+  throttle via `qm set` (as it had done every hour for weeks — see Root
+  cause #4).
+- **15:18–15:20** — guest healthy by every metric: CPU 7–16% of 16 vCPUs,
+  load 1.4, 17 GiB MemAvailable, swap flat at 2.0 GiB, host `sdc` 2–8%
+  utilized. Heavy claude/bwrap sandbox activity (normal workload).
+- **15:19:08** — last journal line the guest ever writes (mid normal
+  traffic, zero kernel distress — not even a hung-task warning).
+- **15:21** — host RRD (pvestatd polling QEMU over QMP once a minute) shows
+  `diskwrite` drop to **exactly 0 and stay 0 for 87 minutes** — not even
+  journal flushes. netout collapses 380K→7K/s. **QEMU keeps answering QMP
+  the whole time** — the process and its main loop are alive; only the
+  block path is dead.
+- **15:21→15:39** — guest CPU (host's view) ramps 11% → ~50% and plateaus:
+  processes progressively piling up behind dead storage (dirty-page
+  writeback stuck → direct reclaim spins). Classic starvation cascade, not
+  a panic (a panic halts or spins flat from t=0).
+- **16:47:42** — QMP socket resets: the wedged QEMU is killed out-of-band
+  (root shell; no PVE task, no snoopy line — shell-builtin `kill`).
+- **16:48:31** — `qmstart` task; guest boots clean on kernel 6.8.0-124
+  (wedged boot ran 6.8.0-117).
+
+## Ruled out (evidence, not vibes)
+
+- **Guest CPU/memory/swap pressure** — healthy at last scrape (Prometheus)
+  and per-minute host RRD.
+- **Host storage** — `pve` thin pool 68% data / 15.5% meta; zero kernel
+  I/O errors on the host all day; `sdc` quiet through the window.
+- **Host-side kill/OOM** — no OOM-killer lines, no segfault, no QEMU crash
+  log; 113 of 114 monitored targets stayed up. Only the devvm died.
+- **Guest kernel panic** — would not keep QMP-visible blockstats frozen at
+  0 while netout ACKs trickle; and the guest kernel logged nothing.
+
+## Root cause
+
+**Class pinned, exact line unprovable** (see below): the devvm's disk I/O
+stalled *inside the QEMU process* — below the guest kernel (all guest I/O
+froze simultaneously with nothing logged) and above host storage (host
+clean, neighbors fine, QEMU main loop responsive). Contributing stack,
+unique to this VM:
+
+1. **`scsihw: lsi`** — the emulated LSI 53C895A (1997 chip, QEMU's legacy
+   default for OSes without virtio drivers). The devvm was the **only VM
+   on the host** running its disk through this path; every healthy
+   neighbor uses `virtio-scsi-pci`. The LSI model is documented as
+   hang-prone under intensive I/O.
+2. **No `iothread`** — all disk emulation ran on QEMU's single main event
+   loop, sharing it with timers and QMP.
+3. **QEMU-level mbps throttle (60/60)** — a token bucket inside QEMU whose
+   queued I/O completes only when its re-arm timer fires.
+4. **Hourly live throttle rewrites** — `apply-mbps-caps.sh`'s idempotency
+   check compared raw config strings, but `qm config` prints keys in its
+   own canonical order, so the check **never matched** and the script
+   re-issued `qm set` (→ live QMP `block_set_io_throttle` against the
+   running QEMU) every hour, 24×/day, for weeks — each poke a chance to
+   race the throttle machinery while queued I/O is in flight. The wedge
+   came 20 min after the 15:01 poke.
+
+## What we could not prove
+
+Whether the stuck queue was the LSI device model, the throttle-group
+timer, or their interaction. The discriminating evidence (QMP
+`query-block`, a stack trace of the QEMU process) existed in RAM at 16:47
+and was destroyed by the recovery kill. If a wedge recurs **autopsy before
+shooting**: `qm guest exec` will fail but `qm monitor`/QMP `query-block`,
+`query-status`, and `gdb -p <pid> -batch -ex 'thread apply all bt'` on the
+kvm process pin it to the line.
+
+## Fixes
+
+| Status | Fix |
+|---|---|
+| shipped (this commit) | `apply-mbps-caps.sh` compares **normalized option sets** — hourly runs are now true no-ops; running VMs' throttle state is no longer rewritten 24×/day. Verified: reordered-key configs compare equal, real drift still triggers `qm set`, post-restart iothread configs compare equal. |
+| staged, awaiting Viktor's cold stop→start | VM 102: `scsihw: virtio-scsi-single` + `scsi0 …,iothread=1,aio=threads` — replaces the LSI path with the paravirt controller all healthy VMs use, moves disk emulation off the main loop, swaps io_uring for boring thread-pool AIO. Guest pre-flight passed (`CONFIG_SCSI_VIRTIO=y` built-in; fstab on LVM dm-uuid/UUID). Must be a **full stop→start** — a guest reboot reuses the old QEMU process. |
+
+## Open follow-ups (discussed 2026-06-11, not yet built)
+
+- `DevvmDown` alert (`up{job="devvm"} == 0 for 3m` → Slack) — closes the
+  90-min detection gap.
+- Freeze forensics: netconsole → pve listener, serial console,
+  `kernel.panic=60`, and a capture-before-kill runbook (above) so any
+  recurrence is pinned, not mourned.
+- The recurring *crawl* class (agent storms → swap-thrash; journald
+  watchdog-killed 3× on 2026-06-10) is a separate failure mode —
+  ssh/tmux sessions remain memory-uncontained by explicit decision
+  (swap-only, 2026-06-10).
+
+## Lessons
+
+- **A VM can die of QEMU-userspace causes that no guest or host kernel log
+  will ever show.** The host's per-VM RRD (pvestatd's QMP polls) is the
+  only witness — `diskwrite=0` with a live QMP socket is the signature.
+- **"Idempotent" reconcilers must prove idempotency against the system's
+  canonical output format**, not against the string they themselves
+  constructed. A compare that never matches turns a safety net into a
+  24×/day fault injector — and its own journal said `updating scsi0`
+  every hour, in plain sight, for weeks.
+- The May-26 mbps caps fixed the sdc-saturation freeze class and
+  introduced this one's trigger surface. Layered mitigations fail in
+  layers — audit what a fix *adds*, not only what it removes.
+- pve host logs are **EEST (UTC+3)**; guest logs are UTC. Every
+  cross-machine correlation in this incident initially looked 3h off.
diff --git a/scripts/apply-mbps-caps.sh b/scripts/apply-mbps-caps.sh
index 39bfb36b..5358f7db 100755
--- a/scripts/apply-mbps-caps.sh
+++ b/scripts/apply-mbps-caps.sh
@@ -27,6 +27,12 @@ TARGETS=(
   "220:scsi0:40:40"      # docker-registry
 )
 
+# Sort a disk spec's comma-separated options so two specs with the same
+# option set but different key order compare equal.
+normalized() {
+  tr ',' '\n' <<<"$1" | LC_ALL=C sort | paste -sd, -
+}
+
 apply_one() {
   local spec="$1"
   local vmid slot rd wr
@@ -49,8 +55,13 @@ apply_one() {
   newvalue="${cleaned},mbps_rd=${rd},mbps_wr=${wr}"
 
   # Skip the qm-set call entirely when state already matches — keeps
-  # journal noise low under the hourly timer.
-  if [[ "$current" == "$newvalue" ]]; then
+  # journal noise low under the hourly timer. Compare option SETS, not raw
+  # strings: `qm config` prints keys in its own canonical order, so a raw
+  # compare never matched and every hourly run re-issued `qm set`, which
+  # live-rewrites the running VM's QEMU throttle state via QMP (implicated
+  # in the 2026-06-11 devvm I/O stall — see
+  # docs/post-mortems/2026-06-11-devvm-qemu-io-stall.md).
+  if [[ "$(normalized "$current")" == "$(normalized "$newvalue")" ]]; then
     echo "vmid $vmid: $slot already at mbps_rd=${rd},mbps_wr=${wr} — no-op"
     return 0
   fi