feat(lambda): Connect CW Live Tail to Lambda functions #6423
Conversation
Allow to "Tail Log Group" from Lambda functions, which will get the configured Log Group for a particular function and start a Live Tail session in that Log Group. Because of the recently updated version of the SDK v2, we can take the configured log group directly from the function's configuration and we don't need an extra call using SDK V3 that was being made in deployedNode.ts
packages/toolkit/.changes/next-release/Feature-48c399ea-d42f-4ea1-8f32-d8bc1486a107.json
Outdated
Show resolved
Hide resolved
| "AWS.command.appBuilder.deploy": "Deploy SAM Application", | ||
| "AWS.command.appBuilder.build": "Build SAM Template", | ||
| "AWS.command.appBuilder.searchLogs": "Search Logs", | ||
| "AWS.command.appBuilder.tailLogs": "Tail Logs", |
There was a problem hiding this comment.
We should probably use the same wording (and actually the same AWS.foo.bar id, instead of multiple) in all places:
aws-toolkit-vscode/packages/core/package.nls.json
Lines 181 to 182 in 8d48a3c
However, not a blocker for this PR. Needs a decision about which wording to use in both places.
There was a problem hiding this comment.
The difference here is that the one you linked shows "when you have a log group selected" (therefore the wording is "tail log group" as in "tail this log group"), while this new one is when you have a function selected. Maybe this could be "Tail function logs" instead of just "Tail logs". The inspiration was the other existing string right above "Search Logs" (with its corresponding "Search Log Group"), which is a similar behavior, but that starts a search instead of using live tail.
Let me know if this difference makes sense or if we should still try to unify the wording.
There was a problem hiding this comment.
The inspiration was the other existing string right above "Search Logs"
Yup I noticed that too, and I like the brevity of "Logs" instead of "Log Group". For now let's leave it as you have in this PR.
| throw ToolkitError.chain( | ||
| err, | ||
| `Unable to fetch logs. Log group for function '${functionConfiguration!.FunctionName}' does not exist. ` + | ||
| 'Invoking your function at least once will create the log group.' |
There was a problem hiding this comment.
Nice hint! Very helpful to surface hints like this.
Thanks for noticing that, eliminating code is very helpful. |
Co-authored-by: Justin M. Keyes <[email protected]>
| } catch (err) { | ||
| if (isError(err as Error, 'ResourceNotFoundException', "LogGroup doesn't exist.")) { | ||
| // If we caught this error, then we know `functionConfiguration` actually has a value | ||
| throw ToolkitError.chain( |
There was a problem hiding this comment.
will we get this issue when we deploy the function for the first time, and hit taillog before invoking it?
There was a problem hiding this comment.
For I feel this should be a valid use case
There was a problem hiding this comment.
Yeah, and that's how the Live Tail API works. It needs an existing log group. We could eventually do something like try once and if it fails we retry (but we might keep retrying forever), but in practice it's better to surface the error, so customers understand it better and retry when needed. Live Tail use case is to "see and filter logs", so it's less likely that people want to use it a lot for a brand new function that doesn't have any invokes yet.
| export function getFunctionLogGroupName(configuration: any) { | ||
| const logGroupPrefix = '/aws/lambda/' | ||
| return configuration.logGroupName || logGroupPrefix + configuration.FunctionName | ||
| return configuration.LoggingConfig?.LogGroup || logGroupPrefix + configuration.FunctionName |
There was a problem hiding this comment.
is this because SDK update?
There was a problem hiding this comment.
It's related. The code that was removed in the other file used to set up this logGroupName variable that wasn't in the original configuration, but here we're just taking it directly from the Lambda Function configuration object that comes from the SDK.
Allow to "Tail Logs" from Lambda functions, which will get the configured Log Group for a particular function and start a Live Tail session in that Log Group. Some code was removed, because with the recently updated version of the SDK v2, we can take the configured log group directly from the function's configuration and we don't need that extra call using SDK V3 that was being made inside `deployedNode.ts` ## Problem To start a CloudWatch Live Tail session, a user needs to know their Log Group name and find it in the AWS explorer. For Lambda users, that might not be easy to find or well known all the time ## Solution We help the Live Tail experience for Lambda customers by allowing them to start a Live Tail session directly from their function in the AWS explorer and in the Application Builder section for deployed functions, without having to search for their Log Group explicitly.
4 participants
Allow to "Tail Logs" from Lambda functions, which will get the configured Log Group for a particular function and start a Live Tail session in that Log Group.
Some code was removed, because with the recently updated version of the SDK v2, we can take the configured log group directly from the function's configuration and we don't need that extra call using SDK V3 that was being made inside
deployedNode.tsProblem
To start a CloudWatch Live Tail session, a user needs to know their Log Group name and find it in the AWS explorer. For Lambda users, that might not be easy to find or well known all the time
Solution
We help the Live Tail experience for Lambda customers by allowing them to start a Live Tail session directly from their function in the AWS explorer and in the Application Builder section for deployed functions, without having to search for their Log Group explicitly.
feature/xbranches will not be squash-merged at release time.