Update / Aktualisierung eScriptorium

[stweil]

Dear all,

currently we run an eScriptorium software version from July/August 2023.

Meanwhile the latest code adds a modern new user interface (UI), but still provides a user specific option for the old "legacy" user interface. It also comes with the latest kraken OCR with faster training. This new code hat some nasty bugs which seem to be fixed now. So maybe it's time to upgrade our installation.

We still have to decide about the time for the upgrade and can take your needs into account. So please comment here or write us directly. As soon as we have planned the upgrade, we will give more information here. Be prepared to make backups of your valuable data with the export function of eScriptorium.

Your
eScriptorium Team at UB Mannheim

[johnlockejrr]

Is there a way to use locally your "build" to test it? I don't mean the official eScriptorium.

[stweil]

Sure, just install https://github.com/UB-Mannheim/escriptorium/tree/es_next instead of the official develop branch.

[johnlockejrr]

How do you run with docker on GPU? Seems nvidia-docker2 is deprecated. Following https://gitlab.com/scripta/escriptorium/-/wikis/docker-install doesn't seem to run on GPU. I normally use a full install of eScriptorium to run on GPU. The celery image is run on GPU as seen below but the tasks are not run on GPU, very strange. Thanks!

570ee13caf16   registry.gitlab.com/scripta/escriptorium:latest   "/usr/src/app/entryp…"   13 hours ago   Up 12 hours                            escriptorium-ub-mannheim-celery-gpu-1

# ps xa
  PID TTY      STAT   TIME COMMAND
    1 ?        Ssl    2:00 /usr/local/bin/python /usr/local/bin/celery -A escriptorium worker -l INFO -E -Ofair --prefetch-multiplier 1 -Q gpu -c 1 --max-tasks-per-child=1
  258 ?        Zs     0:00 [torch_shm_manag] <defunct>
  259 ?        Zs     0:00 [torch_shm_manag] <defunct>
  260 ?        Zs     0:00 [torch_shm_manag] <defunct>
  261 ?        Zs     0:00 [torch_shm_manag] <defunct>
  279 ?        Zs     0:00 [torch_shm_manag] <defunct>
  280 ?        Zs     0:00 [torch_shm_manag] <defunct>
  281 ?        Zs     0:00 [torch_shm_manag] <defunct>
  282 ?        Zs     0:00 [torch_shm_manag] <defunct>
  318 ?        Z      0:02 [celery] <defunct>
  319 ?        Z      0:02 [celery] <defunct>
  320 ?        Z      0:01 [celery] <defunct>
  321 ?        Z      0:01 [celery] <defunct>
  322 ?        Z      0:02 [celery] <defunct>
  323 ?        Z      0:01 [celery] <defunct>
  324 ?        Z      0:02 [celery] <defunct>
  325 ?        Z      0:02 [celery] <defunct>
  344 ?        S      0:00 /usr/local/bin/python /usr/local/bin/celery -A escriptorium worker -l INFO -E -Ofair --prefetch-multiplier 1 -Q gpu -c 1 --max-tasks-per-child=1
  345 pts/0    Ss     0:00 sh
  353 pts/0    R+     0:00 ps xa
# nvidia-smi
Mon Nov  4 11:51:15 2024
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 565.51.01              Driver Version: 565.90         CUDA Version: 12.7     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3060        On  |   00000000:04:00.0 Off |                  N/A |
|  0%   24C    P8             10W /  170W |       0MiB /  12288MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

[stweil]

We use native installations, no Docker.

[johnlockejrr]

Same here, I just thought I'll give a try to docker for a fast check but I get back to native install because I can controll it better.

[johnlockejrr]

What python version and nodejs version do you recommand?

[stweil]

Python 3.11, Nodejs 18.19 (both packages in Debian stable).