Add monitoring of container image deployments #13
Conversation
| @@ -1,4 +1,5 @@ | |||
| module.exports = [ | |||
| require('./waitingpods'), | |||
| require('./longnotready'), | |||
| require('./newdeployment'), | |||
There was a problem hiding this comment.
We're using one tab indentation. Please reindent this line and the monitor file.
|
|
||
| class DeploymentStatus extends EventEmitter{ | ||
| constructor(){ | ||
| super(); |
There was a problem hiding this comment.
If you have nothing but super(); in your constructor, it can be omitted.
There was a problem hiding this comment.
Actually from below, we'll need the constructor to initialize the state store.
| } | ||
|
|
||
| async check(){ | ||
| let containers = await kube.getContainerStatuses(); |
There was a problem hiding this comment.
I think rather than checking for newly deployed containers, we should check for newly deployed pods. This is available in kube.getPods();
| _key: `${item.pod.metadata.name}/${item.name}`, | ||
| }); | ||
|
|
||
| global.image = item.image; |
There was a problem hiding this comment.
If this is stateful then the state should be stored as instance attribute, rather than as a global variable.
|
@nathancashmore would you mind if I hi-jacked this PR and fixed the raised review issues? I'd really appreciate having the bot notify us about new deployments! |
|
@phillipj go for it. |
|
Alrighty, thanks. I'll for sure credit you as the author in the commit! |
|
Updates after having done some initial testing of this; we decided to send a Slack message from our deployment script with a cURL command instead. Looks something like this:
That in combination with the already existing monitors that will tell us about containers having trouble, covers our needs. Bottom line, I was a little quick saying I'll hijack this PR with the goals of getting it merged. Sorry for those who waited in vain. |
|
This may be a little unnecessary for some, this should be conditional with a configuration flag. |
No description provided.