high memory usage

[lukasmrtvy]

Hey,
seems that the exporter consuming up to 2.24GiB memory by default

[egvimo]

Hi @lukasmrtvy,
yes, memory is quite high comparing to others, but I think it comes from the apt_info.py script. A single call to this script takes 5 to 7 seconds. And takes up memory. On my cluster the memory is currently at 450 MB.

Not knowing your setup, my first guess would be to change the scrape interval. Mine is on 2 minutes.

[egvimo]

Hi @lukasmrtvy,
I've replaced the whole web stack of the exporter. Can you try again with the new version?

[lukasmrtvy]

Hi, with 7d scrape interval its 112.8MiB on 0.2.1, previous version was ~50MiB

[egvimo]

I've replaced everything and added cache.close() to the apt_info.py script.

With my previous version the container got OOM killed and restarted (I've set a memory limit to the container). With this new version it's the same.

I have no idea what I can change to prevent the memory increase. It has to be the apt_info.py script, because this is the only common code with these two versions.

[lukasmrtvy]

Btw, see this prometheus-community/node-exporter-textfile-collector-scripts#181

[egvimo]

I've updated the script with the changes from the pull request, but it seems that this was not the cause for the OOM.

[egvimo]

With the latest release it looks like the memory is now stable. @lukasmrtvy can you please test?