If I want to check the logs for my lambda function, I used to log in to the console, go to CloudWatch, find the my lambda (which is made difficult because the search bar doesn’t do substrings), click on my lambda group, and then sort by time, because it doesn’t necessarily sort by time, and then click on the link for the logs. That is exhausting.

Now I use awslogs, which does a lot of the work for me. awslogs groups will list all the existing CloudWatch groups and then awslogs get <group name> will display the latest entries. Of course you pipe it to grep or ag and search for messages. Or you can use --search in the awslogs command. You can also set a start time and/or an end time for the log messages.

This command doesn’t seem to honor the standard AWS_PROFILE environment variable, but you can use --profile <name> at the end of the awslogs command (it doesn’t seem to work anywhere else in the command, unfortunately).

I think of s3cmd as a little Swiss Army knife for all things S3. It does syncing, copying, tree-traversals. It even knows when to use multi-part uploads. Most of my work is in S3 lately, and I use this all the time.

My biggest problem with s3cmd is it doesn’t use AWS’s standard configs but it’s own config file. That makes switching between profiles (something I do a lot) tricky. I ended up making a wrapper script that reads ~/.aws/credentials and takes the given profile name and executes the S3 call. You can see the script here. Usage is:

s3ctl.py <profile_name> <s3cmd command>

so if you are putting fancy-file into s3://my-fancy-bucket/some-folder using your development AWS profile, it’s:

s3ctl.py development put fancy-file  s3://my-fancy-bucket/some-folder