Issue with Supabase Vector on Coolify (Raspberry Pi 5)

[FrancescoBrunoDev]

Hello everyone,
I am trying to get Supabase running on Coolify on a Raspberry Pi 5. The issue is with Supabase Vector, particularly the supabase-vector container, which is returning an error that prevents everything from working.

2024-10-05T16:14:14.841615Z ERROR vector::topology::builder: Configuration error. error=Transform "rest_logs": 
error[E105]: call to undefined function
  ┌─ :4:18
  │
4 │     .timestamp = to_timestamp!(parsed.time)
  │                  ^^^^^^^^^^^^
  │                  │
  │                  undefined function
  │                  did you mean "is_timestamp"?
  │
  = learn more about error code 105 at https://errors.vrl.dev/105
  = see language documentation at https://vrl.dev
  = try your code in the VRL REPL, learn more at https://vrl.dev/examples

Does anyone have any ideas on how to solve this?

Thanks!

[rechbe200163]

Hello, same issue hear just tried to setup everthing up again nothing always the same issue

log from the failing docker:

2024-12-06T18:49:17.876441423Z memory allocation of 5 bytes failed
2024-12-06T18:50:18.348513573Z <jemalloc>: Unsupported system page size
2024-12-06T18:50:18.348579388Z <jemalloc>: Unsupported system page size
2024-12-06T18:50:18.348584443Z memory allocation of 5 bytes failed

have you fixed yours by now?

[FrancescoBrunoDev]

Hi @rechbe200163 ,
Thank you for reminding me about this thread. Yes, I found a solution, and you can find it explained here:
https://pimylifeup.com/raspberry-pi-page-size/

It's a straightforward procedure, and you can revert it without any issues.

It could be necessary to change the to_timestamp thing, though.

There was a pull request on the Supabase repo: https://github.com/supabase/supabase/pull/28931

PS: However, I must tell you that when I tested it on a Raspberry Pi 5, the performance was really poor, likely due to Supabase's AI features.

[hismayilov]

do following change in vector.yml
in
/data/coolify/services/XXXXXXXXXXXX/volumes/logs

•  .timestamp = to_timestamp!(parsed.time)
  

•  .timestamp = parse_timestamp!(parsed.time, format: "%+")