Large file (2.2 GB) gets stuck when trying to upload

[jorgetamayo21]

Describe the bug
I am using the integration with Locust and at the end of a 2 hour and 30 minute test it started uploading but stalled. The file weighs 2.2 GB and I have not found a way to upload it, in the Web UI it was stuck as "item in progress"

To Reproduce
Run a test that generates a 2.2 GB result file

Expected behavior
That the file uploads without problems and can be viewed in reporter

Screenshots

Additional context
upload log at the end of the Locust test
uploading
{"itemId":"33118836-eb29-47fd-a51c-6bdd7893dd79"}
uploading
{"itemId":"3517c50d-9f1d-4138-ae53-f790199becf3"}

[ludeknovy]

Hi @jorgetamayo21 ! Did the container fail? How long was it processing until you declared it as stale? Uploading such a big file is expected to take a long time.

Btw did you try https://jtlreporter.site/docs/integrations/locust#continuous-results-uploading ?
That way the processing should be significantly faster.

[jorgetamayo21]

For about 40 minutes it was processing then the jtl-reporter page crashed, I restarted the machine and it raised the page but the item is still stuck at "item in progress"

I will implement that continuous upload method better, but I will still report the problem to see if I can rescue the results of my test

[jorgetamayo21]

I managed to upload the file using https://github.com/ludeknovy/jtl-reporter/blob/main/scripts/upload_jtl.py, the problem occurred because the machine had only 8 GB of memory and jtl-reporter at the time of processing the file exceeded those 8 GB. I increased it and it processed without problems.

The only detail is that the failed processing still appears stuck as "item in progress", a way should be added to "cancel" files that failed to be processed

[ludeknovy]

The stale item can be changed only directly in the DB for now.

The only mechanism I can think of to delete the failed processing is to set a time threshold (eg: 4 or 8 hours) after which the item would be declared as failed and its status would get changed.

[ludeknovy]

Also, this seems to be the same problem as #259

[jorgetamayo21]

That the failed process expires in 8 hours seems perfect to me.

Regarding the memory problem, I think that if the running machine has enough memory it should not cause problems.

[ludeknovy]

@jorgetamayo21 I've created new FR: #271
I think it makes sense to simply delete the failed/stale test runs, as there's no way to recover from it. And even if I wanted to recover, its integrity was disrupted (it's unknown at what stage of processing the application crashed), thus providing unreliable and possible skewed results.

[ludeknovy]

The problem occurred because the machine had only 8 GB of memory and jtl-reporter at the time of processing the file exceeded those 8 GB.

Was it processing more files at once during that time? The 8GB of RAM should be enough to process one 2.2 GB file. As mentioned, I made some changes to the part of streaming the file into the DB.

Do you think you could remove any sensitive information from your file and share it with me? I don't have such a big file at my disposal, and I would like to perform more tests and see if there's still a problem I could fix with memory management.

[jorgetamayo21]

Give me your email and I'll send you the file

[ludeknovy]

Please send it at ludek @ jtlreporter.site

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[ludeknovy]

Closing this issue, the further steps can be followed here: #259