[Bug]: Desktop stuck on "Connecting to WSL" after update due to renumbered migrations 63-65 (MigrationError) #21936

Before submitting

• I searched existing issues and did not find a duplicate. (Related: [Bug]: Updates from official 0.0.46-preview.2909 fail: migration 61 renumbered (PeerLinks → PullRequestReviewGuides), no upgrade path #21284 for migration renumbering on server; [Bug]: "Connecting to WSL" infinite loading #5967, [Bug]: WSL backend stuck connecting when Windows can't reach the WSL NAT IP #13738 for desktop WSL connection splash).
• I included enough detail to reproduce or investigate the problem.

Area

apps/desktop

Steps to reproduce

1. Install and run T3-Code-0.1.1-preview.20261017.3254-x64.exe with WSL integration enabled.
   This applies migrations up to 65 in ~/.t3/userdata/statev2.sqlite (63: PeerLinks, 64: PeerLinkFence, 65: WaitingLinkRequestIndex).
2. Update/upgrade to T3-Code-0.1.1-nightly.20261019.3373-x64.exe (WSL runtime sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866).
3. Launch T3 Code and connect to WSL.

Expected behavior

The WSL backend process starts up, runs any pending migrations or reconciles renumbered migrations, marks the runtime ready, and connects to the desktop app.

Actual behavior

The desktop UI gets stuck on the "Connecting to WSL" loading screen indefinitely.

Under the hood, the WSL backend child process crashes immediately on launch with exit code 1 because the migrations between the preview build and the nightly build were renumbered/replaced:

• In 0.1.1-preview.20261017.3254: ID 63 was PeerLinks, ID 64 was PeerLinkFence, ID 65 was WaitingLinkRequestIndex.
• In 0.1.1-nightly.20261019.3373: ID 63 is DropUnusedV1Projections, ID 64 is CanonicalizeThreadPullRequestLinks.

Because ID 65 was already recorded, assertV2MigrationsCanRun aborts with MigrationError. The desktop app continuously retries spawning the backend, resulting in an infinite loading loop.

Impact

Blocks work completely

Version or commit

• Upgraded from: T3-Code-0.1.1-preview.20261017.3254
• Upgraded to: T3-Code-0.1.1-nightly.20261019.3373
• WSL Runtime: sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866

Environment

Windows 11, WSL2 (Ubuntu), T3 Code Desktop 0.1.1-nightly.20261019.3373

Logs or stack traces

From %USERPROFILE%\.t3\userdata\logs\server-child.log:

{"message":"backend child process failure output start","level":"ERROR","timestamp":"2026-10-19T21:10:10.471Z","annotations":{"component":"desktop-backend-child","runId":"f6a6d11d3856","instanceId":"primary","phase":"START","details":"pid=8020 port=3774 cwd=C:\\Users\\Hamza"},"spans":{},"fiberId":"#backend-child"}
{"message":"backend child process output","level":"INFO","timestamp":"2026-10-19T21:10:10.472Z","annotations":{"component":"desktop-backend-child","runId":"f6a6d11d3856","instanceId":"primary","stream":"stdout","text":"[02:40:10.433] ERROR (#1): MigrationError: This database's migration history would skip required schema migrations: 63_DropUnusedV1Projections (recorded: PeerLinks), 64_CanonicalizeThreadPullRequestLinks (recorded: PeerLinkFence). The migrator skips every id up to the highest recorded id (65), regardless of name. It was probably written by a different build. Start the build that wrote it, or report this migration history so an upgrade path can be added.\n    at file:///home/hamza/.t3/wsl-runtime/sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866/t3:116153:17\n    at assertV2MigrationsCanRun (file:///home/hamza/.t3/wsl-runtime/sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866/t3:116162:26)\n    at assertV2MigrationsCanRun (definition) (file:///home/hamza/.t3/wsl-runtime/sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866/t3:116138:29)\n    at runMigrations (file:///home/hamza/.t3/wsl-runtime/sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866/t3:116243:10)\n    at runMigrations (definition) (file:///home/hamza/.t3/wsl-runtime/sha256-3ab3de9154fd1724be4800fa01d3df0906ffa35fa3e7a7ed1c4d7110b9c7a866/t3:116159:18)"},"spans":{},"fiberId":"#backend-child"}
{"message":"backend child process failure output end","level":"ERROR","timestamp":"2026-10-19T21:10:10.473Z","annotations":{"component":"desktop-backend-child","runId":"f6a6d11d3856","instanceId":"primary","phase":"END","details":"pid=8020 code=1"},"spans":{},"fiberId":"#backend-child"}

Recorded migration history before manual intervention in ~/.t3/userdata/statev2.sqlite:

SELECT migration_id, name FROM effect_sql_migrations ORDER BY migration_id DESC LIMIT 5;
-- 65 | WaitingLinkRequestIndex
-- 64 | PeerLinkFence
-- 63 | PeerLinks
-- 62 | ThreadFailureLookupIndex
-- 61 | PullRequestReviewGuides

Workaround

Close T3 Code and delete conflicting migration rows >= 63 from effect_sql_migrations in both ~/.t3/userdata/statev2.sqlite (WSL side) and %USERPROFILE%\.t3\userdata\statev2.sqlite (Windows side):

sqlite3 ~/.t3/userdata/statev2.sqlite "DELETE FROM effect_sql_migrations WHERE migration_id >= 63;"
sqlite3 /mnt/c/Users/<user>/.t3/userdata/statev2.sqlite "DELETE FROM effect_sql_migrations WHERE migration_id >= 63;"

Once the recorded migration ceiling is reset to 62 (ThreadFailureLookupIndex), restarting T3 Code successfully applies 063_DropUnusedV1Projections and 064_CanonicalizeThreadPullRequestLinks, and the WSL backend connects normally.