search

virocon-organization / viroconweb

A software to compute environmental contours. Django web-application package.
MIT License
2 stars 1 forks source link

Heroku Error R14 (Memory quota exceeded) #180

Open ahaselsteiner opened 6 years ago

ahaselsteiner commented 6 years ago

I'm submitting a ...

Expected behavior

How should it work (with the bug fixed or the feature implemented)? Memory should be sufficient.

Screenshot

Actual behavior

How does it currently work (with the bug causing problems or without the feature)? The 512 MB memory is exceeded and Heroku responses with the Error R14 (Memory quota exceeded). The user is displayed "Application error".

Screenshot image

Steps to reproduce the problem (how to see the actual behavior)

  1. Upload the test measurement file.
  2. Fit 4 times after each other this model to it: Wbl,LN(power, exp), bin width = 0.5
  3. The third or fourth time you should encounter the R14 error.
ahaselsteiner commented 6 years ago

I tried to get the error by fitting a Wbl,LN(power, exp), bin width = 0.1. This might have exceeded the 512 MB ram by performing this fit only once. However, it timed out (with MAX_COMPUTING_TIME = 15 s). This lead me to one possible solution to the problem: If we decrease the MAX_TIME_OUT to maybe something like 5 s, a bin width of 0.5 would also time out and the memory would not be exceeded.

The MAX_COMPUTING_TIME depends on the the RUN_MODE (production vs local). Consequently, users could do these memory-intensive calculatons only locally. I will give this a try.

ahaselsteiner commented 6 years ago

For some reason the calculations today take much longer than usually.

Fitting a WBL,WBL(none,none) model to the test data set even exceeded 5 s computing time.

image

image