We have an application (off the shelf, with a time-based and feature-based licence key) running on a Windows Server EC2 instance. This application creates logs which are stored in the following directory: C:\ProgramData\MyApp\v4\Logs. In this folder, there are many files with names that are {guid}.txt.
We want to feed these to Cloudwatch. We created the config.json but I wanted to ask:
Does the file_path in the config.json have to be pointing to one and only one file? Is it possible to point to a folder?
{
"logs": {
"logs_collected": {
"files": {
"collect_list": [
{
"file_path": "C:\\ProgramData\\MyApp\\v4\\Logs\\logs.txt",
"log_group_name": "Logs-Myapp",
"log_stream_name": "{instance_id}-Myapp",
"retention_in_days": 180
}
]
}
}
}
}
I tried pointing to a folder but I am not getting any results.
The file_path inside the collect_list can only point to files. However it does allow for wildards inside the file name to select a range of files. So if your files are always named as {guid}.txt you can use the following pattern.
{
"logs": {
"logs_collected": {
"files": {
"collect_list": [
{
"file_path": "C:\\ProgramData\\MyApp\\v4\\Logs\\*.txt",
"log_group_name": "Logs-Myapp",
"log_stream_name": "{instance_id}-Myapp",
"retention_in_days": 180
}
]
}
}
}
}
The official AWS documentation page for the logging section of the cloud watch agent also describes that you can use a super asterisk (**) within file paths. In theory that could mean that you can also try the following pattern:
"file_path": "C:\\ProgramData\\MyApp\\v4\\Logs\\**"
However, I’ve never tried that and the pattern might be a bit too permissive.