ECONRESET while upload image files using adobe/aio-lib-files
- Dominant language
- JavaScript
- Stars
- 95
- Forks
- 45
- PR merge metrics
- No merged PRs in 30d
Description
### Expected Behavior
Asset compute uses https://github.com/adobe/aio-lib-files to generate presigned url and upload image files. We are seeing large number of ECONRESET errors when trying to upload image files. This results in the request to be stuck for around 3 mins before the upload completes. This happened to 0.1% of our requests.
SLACK Thread: https://cq-dev.slack.com/archives/C013UDBFBGB/p1687823977186229
### Actual Behaviour
ECONRESET should not happen. The client should not be
### Reproduce Scenario (including but not limited to)
uploading image
#### Steps to Reproduce
```
"domain": "windows.net",
"duration": 164275.366014,
"durationBlocked": 0.34779,
"error": 1,
"errorCode": "ECONNRESET",
"errorMessage": "read ECONNRESET",
"host": "aiocnastorageprod.blob.core.windows.net",
"method": "PUT",
"path": "/60766455e9a22e79b92a72fda4d35419/207e9ffc-8957-4f52-8022-3194df5a81e1/1687170266325/file.tmp?sv=...",
"port": 443,
"protocol": "https:",
"sourceMimetype": "image/jpeg",
"sourceName": "Test_Picture_10MB - Kopie (3).jpg",
"sourceSize": 12327389,
"timestamp": 1687170430612,
"transactionId": "6p8l42abgx0Z8Jb8nFuv8qkvy5hHGfsE",
"url": "https://aiocnastorageprod.blob.core.windows.net/60766455e9a22e79b92a72fda4d35419/207e9ffc-8957-4f..."
```
#### Environment Info
```
System:
OS: macOS 13.2
CPU: (10) x64 Apple M1 Max
Memory: 184.51 MB / 32.00 GB
Shell: 5.8.1 - /bin/zsh
Binaries:
Node: 14.19.0 - ~/.nvm/versions/node/v14.19.0/bin/node
Yarn: 1.22.19 - ~/.nvm/versions/node/v14.19.0/bin/yarn
npm: 6.14.16 - ~/.nvm/versions/node/v14.19.0/bin/npm
Virtualization:
Docker: 20.10.17 - /usr/local/bin/docker
npmGlobalPackages:
@adobe/aio-cli: Not Found
```
#### Sample Code that illustrates the problem
https://github.com/adobe/asset-compute-pipeline/blob/master/lib/storage/temporary-cloud-storage.js#L115
#### Logs taken while reproducing problem
Contributor guide
Assessment
This issue has not been assessed yet.