Description
Large(up to 150 megs) supporting claim evidence file uploads are failing out with timeout errors.
The large files are split and uploaded to EVSS in 50 meg chunks after the claim is successfully submitted.
Not all file hit the issue, I think I had a quick look a few months ago, but I really was not all that "up" on the upload code..but at the time, I noticed at least some of them looked like the read from the temp file location in amazon s3 was timing out and or running the sideki process out of memory on occasion.
These files are uploaded to the VA via EVSS.
Notes-Ideas:
The files are stored in s3 to start with, kind of a temp storage while the user is in the middle of creating their claim.
Once the claim is submitted, we read the entire file into memory(s3 has limited support for reading the file in smaller chunks) so this might be the part of the issue, we might be running out of memory.
Also, look at the timeout settings, and try to figure out the times between the chunk send request and the response from evss. I saw a thread at some point that seemed like someone was implying the response was taking a while to make it back to vets-api and the delay was accumulating due to the chunked multiple upload calls
Impact: we continue to see about 50 files per month hit this issue
Activity