-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Task run_osmtgmod fails #258
Comments
Update: please don't act here for now, because it might be related to a |
I checked again and I'm quite sure that this osmtgmod error is independent from insufficient disk space. When reading this part of the error message |
@ClaraBuettner will add manual setting of the SRID for this geom. Additionally, we try to understand why the workflow ran successfully on the Flensburg server but not RLI server by comparing python package versions. |
The following versions of packages which might be related to this are installed on the server in Flensburg: |
These are our package versions SQLAlchemy==1.3.24 The only difference (highlighted) is in the package version of SQLAlchemy-Utils. |
Thanks to the fix of manually set SRID #263 , the error does not occur anymore.
I remember the @ClaraBuettner @ulfmueller do you have an idea why the task still fails? |
The above error might be related to a to small |
@gplssm If a different |
I don't know why this task still fails, installing java always solved this issue in my case.
I didn't changed the |
Would be great if you could check this. Please use |
Thank you! The output of |
Thanks! |
@gplssm's trial to increase I created everything from scratch on the same server and face the very same problem after 1 day. Taking this to extreme with rather ridiculous values I increased the Could the table locking problem be related to concurrent processes, but there's no extra multiprocessing in the osmTGmod task, right @ClaraBuettner ? As we dunno how to tackle this further, I have a request to you @gnn: if you can spare some time, could you please set up eGon-data on one of your machines to see whether u run into the same transaction jungle? Relevant part of the log (last entry seems to be running since startup..):
|
Just as an information, in my case a run for Germany needs ~1.5 days (6 tasks in parallel, 1 process per task). osmTGmod finishes after ~6 hours.
As far as I know multiprocessing is only used in the osm import (not the one for osmtgmod) and some pandas operations. In #269 you can also limit the number of parallel processes per task. |
Thanks for your quick reply! |
Now this problem also occurs with the SH test region, not only for entire Germany. And on 3 different machines including your server in Flensburg with a fresh venv and current dev. @ClaraBuettner @IlkaCu @ulfmueller @gnn @ClaraBuettner, do you dare to re-run this task in your DE version cause I don't.. Log:
PS: Again, increasing |
The task run successfully about 10 days ago on the current dev branch. I double-checked that the task was running completely.
I don't want to re-run the task, but I can create a new database in test mode. And I will create a backup of the osmtgmod results for Germany. |
Thanks for double checking this!
Thank you. I'd think that there's not too much to screw up during setup?!
I'm curious whether you chose a different way for your setup? |
I choose only 6 parallel jobs since some tasks also have an internal multi-threading. I'm not sure if that might have caused your problems. |
Unfortunately not, I also tried with 2 jobs. I'm curious if your trial will work out....oh, this suspense is killing me... :) |
I created a new instance of egon-data (in test mode) on the server, now I also run into the error. |
Okay, now I also get this error on my local machine in all environments that worked before. I also created a new one, and osmtgmod fails with the same error. |
Damn it! How on earth is that possible? @ClaraBuettner should we have a meeting on this to brainstorm the possible reasons? |
Thanks for testing! Damn, I was hoping that I did something wrong... |
This is the exact same error that occurs when java was not installed (see #206). I assumed that there is a problem with the java installation. and tried to uninstall and re-install java, but it didn't help. Then I had a look into my database and noticed that all tables in the public schema which are used as an input for osmtgmod are empty. In addition the The function eGon-data/src/egon/data/processing/osmtgmod/__init__.py Lines 144 to 163 in fb1b84c
I replaced the subprocess.popen with the egon-data internal subprocess function to solve the bug described there. When I did this, the task
So there is a problem with java, osmosis (which imports the input data for osmtgmod) fails and we just didn't saw this error message before. |
I finally found the problem and it is very easy to solve. The pbf file is stored in another directory since migrating the osm-download to datasets. When I change the path, it is working. I will prepare a PR which solves this. The old file is not deleted in directories which were created before. That is why it is not failing in existing instances of egon-data. (On my local machine I had the smart idea to clean up my folders and deleted all old files, which is why the error also occurred in my former running instances.) |
🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 🎉 |
Fix path to pbf file in osmtgmod_osm_file
While SH ran successfully, DE did not. |
I remember that the parallel task |
I found the problem, see #344 |
After fixing #344 |
The task
run_osmtmod
fails, but only with--dataset-boundary=Everything
.The text was updated successfully, but these errors were encountered: