TheAnachronism/docspell
1
0
Fork 0
mirror of https://github.com/TheAnachronism/docspell.git synced 2025-03-28 17:55:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update docs on rpi regarding requirements for joex

Browse Source
This commit is contained in:
Eike Kettner 2020-09-25 21:58:02 +02:00
parent 616284c167
commit a36c499fb1
2 changed files with 23 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
nix/configuration-test.nix
View File

@ -35,7 +35,7 @@ in
waitForTarget = "solr-init.target";
bind.address = "0.0.0.0";
base-url = "http://localhost:7878";
jvmArgs = [ "-J-Xmx2g" ];
jvmArgs = [ "-J-Xmx1536M" ];
inherit full-text-search;
};
services.docspell-restserver = {

35
website/site/content/docs/install/rpi.md
View File

@ -19,19 +19,28 @@ work is done by the joex components.
## Joex
Running the joex component on the Raspberry Pi is possible, but will
result in long processing times for OCR. Files that don't require OCR
are no problem.
result in long processing times for OCR and text analysis. The board
should provide 4G of RAM (like the current RPi4), especially if also a
database and solr are running next to it. I recommend to give joex a
heap of 1.5G (`-J-Xmx1536M`). You should also set the joex pool size
to 1.
Tested on a RPi model 3 (4 cores, 1G RAM) processing a PDF (scanned
with 300dpi) with two pages took 9:52. You can speed it up
considerably by uninstalling the `unpaper` command, because this step
takes quite long. This, of course, reduces the quality of OCR. But
without `unpaper` the same sample pdf was then processed in 1:24, a
speedup of 8 minutes.
When joex processes the first file, some models are built loaded into
memory which can take a while. Subsequent processing times are faster
then.
You should limit the joex pool size to 1 and, depending on your model
and the amount of RAM, set a heap size of at least 500M
(`-J-Xmx500M`).
An example: on this [UP
board](https://up-board.org/up/specifications/) with an Intel Atom
x5-Z8350 CPU (@1.44Ghz) and 4G RAM, a scanned (300dpi) pdf file with 6
pages took *3:20 min* to process. This board also runs the SOLR and a
postgresql database.
For personal setups, when you don't need the processing results asap,
this can work well enough.
The same file was processed in 55s on a qemu virtual machine on my i7
notebook, using 1 CPU and 4G RAM (and identical config for joex). The
virtual machine only had to host docspell (joex and restserver, but
the restserver is very lightweight).
The learning task for text classification can also use high amount of
memory, but this depends on the amount of data you have in docspell.
If you encounter problems here, you can set the maximum amount of
items to consider in the collective settings page.