Clean timescale/timescaledb-ha:pg17.5-ts2.20.2 (linux/arm64) - "TimescaleDB Background Worker Launcher" Illegal instruction

[schacka]

Docker Image timescale/timescaledb-ha:pg17.5-ts2.20.2 does not start successfully on my RPi4 (linux/arm64). Current pull and empty data directory. timescale/timescaledb-ha:pg17.5-ts2.20.1 (linux/arm64) works fine. On my server there are no errors with timescale/timescaledb-ha:pg17.5-ts2.20.2 (linux/amd64 ).
What could be the problem?

Docker Compose:

services:
   db:
    image: timescale/timescaledb-ha:pg17
    container_name: ${POSTGRES_HOST}
    hostname: ${POSTGRES_HOST}
    restart: unless-stopped
    environment:
        TZ: ${TZ}
        POSTGRES_USER: ${POSTGRES_USER}
        POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
        POSTGRES_DB: ${POSTGRES_DB}
        LANG: ${LANG}
        POSTGRES_INITDB_ARGS: ${POSTGRES_INITDB_ARGS}
        TS_TUNE_MEMORY: ${TS_TUNE_MEMORY}
        TS_TUNE_NUM_CPUS: ${TS_TUNE_NUM_CPUS}
        TS_TUNE_MAX_BG_WORKERS: ${TS_TUNE_MAX_BG_WORKERS}
        TS_TUNE_MAX_CONNS: ${TS_TUNE_MAX_CONNS}
        TIMESCALEDB_TELEMETRY: ${TIMESCALEDB_TELEMETRY}
        # Uncomment from here after first init
        PGHOSTADDR: 127.0.0.1 
        PGPORT: 5432
        PGDATABASE: ${POSTGRES_DB}
        PGUSER: ${POSTGRES_USER}
        PGPASSWORD: ${POSTGRES_PASSWORD}
    volumes:
      - testdb_data:/home/postgres/pgdata/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U $$POSTGRES_USER -d $$POSTGRES_DB"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 60s

Docker .env:

TZ: Europe/Berlin
POSTGRES_HOST: testdb
POSTGRES_DB: test
POSTGRES_USER: test
POSTGRES_PASSWORD: test
LANG: de_DE.utf8
POSTGRES_INITDB_ARGS: "--locale-provider=icu --icu-locale=de-DE"
TS_TUNE_MEMORY: 512MB
TS_TUNE_NUM_CPUS: 2
TS_TUNE_MAX_BG_WORKERS: 16
TS_TUNE_MAX_CONNS: 150
TIMESCALEDB_TELEMETRY: off

Docker Logs:

2025-06-10 07:26:13.667 CEST [1827] LOG:  checkpoint complete: wrote 3 buffers (0.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.011 s, sync=0.003 s, total=0.038 s; sync files=2, longest=0.002 s, average=0.002 s; distance=0 kB, estimate=0 kB; lsn=0/190BBB0, redo lsn=0/190BBB0
2025-06-10 07:26:13.698 CEST [1] LOG:  database system is ready to accept connections
2025-06-10 07:26:14.250 CEST [1] LOG:  background worker "TimescaleDB Background Worker Launcher" (PID 1831) was terminated by signal 4: Illegal instruction
2025-06-10 07:26:14.250 CEST [1] LOG:  terminating any other active server processes
2025-06-10 07:26:14.260 CEST [1] LOG:  all server processes terminated; reinitializing
2025-06-10 07:26:14.339 CEST [1833] LOG:  database system was interrupted; last known up at 2025-06-10 07:26:13 CEST
2025-06-10 07:26:14.434 CEST [1833] LOG:  database system was not properly shut down; automatic recovery in progress
2025-06-10 07:26:14.456 CEST [1833] LOG:  invalid record length at 0/190BC28: expected at least 24, got 0
2025-06-10 07:26:14.456 CEST [1833] LOG:  redo is not required
2025-06-10 07:26:14.494 CEST [1834] LOG:  checkpoint starting: end-of-recovery immediate wait
2025-06-10 07:26:14.527 CEST [1834] LOG:  checkpoint complete: wrote 3 buffers (0.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.011 s, sync=0.004 s, total=0.039 s; sync files=2, longest=0.002 s, average=0.002 s; distance=0 kB, estimate=0 kB; lsn=0/190BC28, redo lsn=0/190BC28
2025-06-10 07:26:14.559 CEST [1] LOG:  database system is ready to accept connections
2025-06-10 07:26:15.137 CEST [1] LOG:  background worker "TimescaleDB Background Worker Launcher" (PID 1838) was terminated by signal 4: Illegal instruction
<Repeat of the above>

[tazmdevil]

I have the same problem with timescale/timescaledb-ha:pg16.9. The ts2.20.2 and ts2.20.3 images produce the same logs. The ts2.20.1 image works. Is the RPI4 processor no supported?

[Starfoxfs]

Same here:

version: "3.0"
name: "TimeScaleDB"
services:
timescaledb:
image: timescale/timescaledb-ha:pg17
container_name: timescaledb
ports:
- 5432:5432
environment:
POSTGRES_USER: user
POSTGRES_PASSWORD: pass
volumes:
- /home/docker/timescale/data:/pgdata
logging:
driver: json-file
restart: unless-stopped

2025-07-13 16:52:14.673 UTC [1] LOG: terminating any other active server processes
2025-07-13 16:52:14.676 UTC [1] LOG: all server processes terminated; reinitializing
2025-07-13 16:52:14.762 UTC [133] LOG: database system was interrupted; last known up at 2025-07-13 16:52:13 UTC
2025-07-13 16:52:14.927 UTC [133] LOG: database system was not properly shut down; automatic recovery in progress
2025-07-13 16:52:14.931 UTC [133] LOG: invalid record length at 0/1B227C8: expected at least 24, got 0
2025-07-13 16:52:14.931 UTC [133] LOG: redo is not required
2025-07-13 16:52:14.938 UTC [134] LOG: checkpoint starting: end-of-recovery immediate wait
2025-07-13 16:52:14.954 UTC [134] LOG: checkpoint complete: wrote 3 buffers (0.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.006 s, sync=0.002 s, total=0.019 s; sync files=2, longest=0.002 s, average=0.001 s; distance=0 kB, estimate=0 kB; lsn=0/1B227C8, redo lsn=0/1B227C8
2025-07-13 16:52:14.965 UTC [1] LOG: database system is ready to accept connections
2025-07-13 16:52:15.979 UTC [1] LOG: background worker "TimescaleDB Background Worker Launcher" (PID 138) was terminated by signal 4: Illegal instruction

[Horner14]

Same here. Works fine on my Pi5

[Starfoxfs]

I have now Docker latest PG17 installed on RPI4 without any issues

[schacka]

I have now Docker latest PG17 installed on RPI4 without any issues

@Starfoxfs : PG17 or PG 18? Just pulled newest PG17 Docker on RPi4, the old poblem still exists.
"background worker "TimescaleDB Background Worker Launcher" (PID 62) was terminated by signal 4: Illegal instruction"

[Starfoxfs]

PG17. But i use now this Docker Image: timescale/timescaledb:latest-pg17

[schacka]

@Starfoxfs: Ok, but then please don't confuse everyone saying, you have no issues. The error is about timescaledb-ha and not timescaledb. That's why i posted the issue using this repo. I use timescaledb-ha for HomeAssistant with PostGIS. So a working timescaledb (aka TimescaleDB Light) doesn't help me much.

[kernstock]

We are seeing the same problem. And in our case using the timescale/timescaledb-ha:2.20.1-pg16 image helped as a workaround.

[Starfoxfs]

@schacka yes your right, i have used the wrong Timescaledb. I have used Timescaledb-ha what i dont need. HA is for Homeassistant.

The normal one without Homeassistant works for me so my post here is in the wrong Thread.

[kernstock]

@schacka yes your right, i have used the wrong Timescaledb. I have used Timescaledb-ha what i dont need. HA is for Homeassistant.

Just for the sake of completeness, HA is for "High Availability" here.

[schacka]

@kernstock :) ... yapp, all the problems started with ts2.20.2. I'm just wondering why the ARM image is not being fixed or removed, if it hasn't been working since June anyway...

[beimoe]

I got the same error messages on my RPI4 arm64. Rolling back to timescaledb-2-loader-postgresql-15=2.20.1debian12 and timescaledb-2-postgresql-15=2.20.1debian12 fixed it. I am using repo https://packagecloud.io/timescale/timescaledb/debian bookworm/main

[razorree]

I rolled back on my RPi4 to

timescaledb-2-loader-postgresql-18/trixie,now 2.23.0~debian13 arm64 [installed,upgradable to: 2.24.0~debian13-1801]
timescaledb-2-postgresql-18/trixie,now 2.23.0~debian13 arm64 [installed,upgradable to: 2.24.0~debian13-1801]
timescaledb-tools/trixie,now 0.18.1~debian13 arm64 [installed,automatic]

and it works again

[graveland]

We're looking into it. The build options changed and it apparently required a newer arm (arm v8.2 vs v8 or something like that). I think it should still be possible with pis, but the build options will need another tweak.