[BUG]: Database fails to init after update

[GyulaSipos]

To Reproduce
I updated my dependencies from pglite 0.3.15 and pglite-sync 0.4.0 to pglite 0.4.1 and pglite-sync 0.5.1.
No other modifications where made in the project. Run it in debug. Chrome dev console is filled with errors, database fails to init.
I even tried deleting my idb (where pglite persists), did not help.

Logs

tec: shmctl: no such segment 666
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.813 UTC [42] DEBUG:  invoking IpcMemoryCreate(size=3981312)
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.815 UTC [42] DEBUG:  dynamic shared memory system will support 304 segments
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.815 UTC [42] DEBUG:  created dynamic shared memory control segment 1049979052 (7308 bytes)
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.816 UTC [42] DEBUG:  InitPostgres
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.817 UTC [42] LOG:  database system was interrupted; last known up at 2026-03-22 07:41:16 UTC
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.817 UTC [42] DEBUG:  removing all temporary WAL segments
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  checkpoint record is at 0/3FA0760
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  redo record is at 0/3FA0760; shutdown true
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  next transaction ID: 757; next OID: 40960
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  next MultiXactId: 1; next MultiXactOffset: 0
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  oldest unfrozen transaction ID: 730, in database 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  oldest MultiXactId: 1, in database 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  commit timestamp Xid oldest/newest: 0/0
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] LOG:  database system was not properly shut down; automatic recovery in progress
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  transaction ID wrap limit is 2147484377, limited by database with OID 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  MultiXactId wrap limit is 2147483648, limited by database with OID 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  starting up replication slots
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  xmin required by slots: data 0, catalog 0
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  starting up replication origin progress state
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  didn't need to unlink permanent stats file "pg_stat/pgstat.stat" - didn't exist
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.818 UTC [42] DEBUG:  resetting unlogged relations: cleanup 1 init 0
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.819 UTC [42] LOG:  redo starts at 0/3FA07D8
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.822 UTC [42] LOG:  invalid record length at 0/3FCE1C8: expected at least 24, got 0
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.822 UTC [42] LOG:  redo done at 0/3FCDED8 system usage: CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.822 UTC [42] DEBUG:  resetting unlogged relations: cleanup 0 init 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.823 UTC [42] DEBUG:  MultiXactId wrap limit is 2147483648, limited by database with OID 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.823 UTC [42] DEBUG:  MultiXact member stop limit is now 4294914944 based on MultiXact 1
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.823 UTC [42] DEBUG:  performing replication slot checkpoint
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.824 UTC [42] DEBUG:  attempting to remove WAL segments older than log file 00000000000000000000003E
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.825 UTC [42] DEBUG:  cache state reset
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.827 UTC [42] FATAL:  role "web_user" does not exist
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.827 UTC [42] DEBUG:  shmem_exit(1): 6 before_shmem_exit callbacks to make
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.827 UTC [42] NOTICE:  shutting down
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.827 UTC [42] DEBUG:  performing replication slot checkpoint
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  attempting to remove WAL segments older than log file 00000000000000000000003E
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  SlruScanDirectory invoking callback on pg_subtrans/0000
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  shmem_exit(1): 12 on_shmem_exit callbacks to make
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  cleaning up orphaned dynamic shared memory with ID 3330523956
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  cleaning up dynamic shared memory control segment with ID 1049979052
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  proc_exit(1): 2 callbacks to make
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] NOTICE:  database system is shut down
et @ index.js?v=3d727c28:3
index.js?v=3d727c28:3 2026-03-22 07:44:11.828 UTC [42] DEBUG:  exit(1)
et @ index.js?v=3d727c28:3
138index.js?v=3d727c28:3 Uncaught (in promise) Error: PGlite failed to initialize properly
    at q.nt (index.js?v=3d727c28:3:283062)
    at q.tt (index.js?v=3d727c28:3:280141)
    at async q.create (index.js?v=3d727c28:3:272661)
    at async init (db_worker.ts:22:20)
5318index.js?v=3d727c28:1 Uncaught (in promise) Error: PGlite failed to initialize properly
    at q.nt (index.js?v=3d727c28:3:283062)
    at q.tt (index.js?v=3d727c28:3:280141)
    at async q.create (index.js?v=3d727c28:3:272661)
    at async init

Details

• PGlite version: 0.4.1
• using any extensions? pglite-sync 0.5.1
• OS version mac
• up to date Arc browser (chromium)

[tdrz]

index.js?v=3d727c28:3 2026-03-22 07:44:11.827 UTC [42] FATAL: role "web_user" does not exist

Interesting. Could you please try in a private/incognito window?

Also, try our online REPL, it does an init every time.

[jljorgenson18]

I also ran into this and I believe it was specific to running it via the worker module. I can get more information for you but I had to roll back to 0.3. I also made sure I was starting with a clean slate and not trying to load in previous DB files.

[tdrz]

I can get more information for you

@jljorgenson18 Yes, please do!

[xiaoneng]

Same issue after upgrading from 0.3.1 to 0.4.1. Reverted to 0.3.1 everything ok.

error: Error: PGlite failed to initialize properly

Database query error: Error: PGlite failed to initialize properly
at q.nt (file:///app/node_modules/@electric-sql/pglite/dist/index.js:3:283062)
at q.tt (file:///app/node_modules/@electric-sql/pglite/dist/index.js:3:280141)
at async q.create (file:///app/node_modules/@electric-sql/pglite/dist/index.js:3:272661)
at async getDB (file:///app/out/main/index.js:22:15)
at async file:///app/out/main/index.js:69:18
at async Session. (node:electron/js2c/browser_init:2:107107)

[tdrz]

Same issue after upgrading from 0.3.1 to 0.4.1. Reverted to 0.3.1 everything ok.

@xiaoneng I need more details. Please provide, if possible, a minimal reproducible code sample.

[tdrz]

at async Session. (node:electron/js2c/browser_init:2:107107)

Are you using an electron app?

[xiaoneng]

at async Session. (node:electron/js2c/browser_init:2:107107)

Are you using an electron app?

Yes

[xiaoneng]

Errors @ getDB:

ipcMain.handle("db:query", async (_, filePath, sql, params) => {
	try {
		return await (await getDB(filePath)).query(sql, params);
	} catch (error) {
		console.error("Database query error:", error);
		return {
			error: true,
			message: error instanceof Error ? error.message : String(error),
			name: error instanceof Error ? error.name : "Unknown Error"
		};
	}
});

here is getDB function:

async function getDB(filePath) {
	if (dbInstance) return dbInstance;
	dbInstance = await PGlite.create({
		dataDir: filePath,
		debug: 0,
		extensions: {
			live,
			uuid_ossp,
			vector
		}
	});
	return dbInstance;
}

[tdrz]

@xiaoneng So you're using it in a web worker iiuc

[tdrz]

dataDir: filePath,

@xiaoneng is the filePath already populated? Could you move it temporarily somewhere else and try with an empty one?

[TeeAaTeeUu]

For me the issue was a bit of misunderstanding how the database config in PGlite.create({ database: 'example' }) behaved. Removing that solved the problem for me, as I was only running this in its default dataDir: 'memory://' mode. Removing the database property from config helped.

Another migration I needed to do, was to init await db.exec(`CREATE USER postgres`) in my test-runner setup, as my schema was referring to that default user.

[tdrz]

@TeeAaTeeUu Thank you!

Another migration I needed to do, was to init await db.exec(`CREATE USER postgres`) in my test-runner setup, as my schema was referring to that default user.

Adding postgres user and making it the default is on my todo list.

[xiaoneng]

dataDir: filePath,

@xiaoneng is the filePath already populated? Could you move it temporarily somewhere else and try with an empty one?

filePath was already populated. I removed, recreated db and 0.4.1 and 0.4.2 BOTH work.

FYI. IF db not recreated, 0.4.2 also failed with same error but I'm fine with recreating db each upgrade now that I'm aware.

[tdrz]

Thank you @xiaoneng !

FYI. IF db not recreated, 0.4.2 also failed with same error but I'm fine with recreating db each upgrade now that I'm aware.

Recreating or migrating as described on our docs should only be needed between minor or major versions. If it didn't work if you started with 0.4.1 and then used 0.4.2 that's an issue. But from 0.3.x to 0.4.x you should recreate or migrate.