Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
Datadog Feedback 24/01/2014

@datadoghq - Overall well done! Here's some feedback:

Signup/User process

  • when someone invites you to their datadog/team, you get an email but because you are already registered you can't join their team. Especially annoying for freelancers that work for multiple customers but don't have an email account at the customer site
  • disable a user is scary (will this delete it, can I re-create it?); a warning or ? next to it would help for fast understanding
  • if the user you invite does not have it's gmail activated yet (google for apps), when the email is added it's send out (but doesn't reach it of course). There is no way of re-sending the invite from the admin side
  • the forcing of registering at least one agent/integration is really annoying. if you signup through an ipad or the person signing up (like billing) does not have any way to go past that screen. So you wonder , am I registered or not? In our case the person installed the mac agent from it's laptop to get it going. No need for that imho
  • when signing up , I wondered about the 100 hosts limit: on EC2 you constantly set/recreate/destroy new machines; how is that counted? Can I delete a host? I could not totally understand it
  • not sure what happens with your metrics/data, can you export it if you stop the account?
  • is there a test/dev only account? I'd like to recommend it to customers and do some development stuff; do I just get a free account after the trial expires with specific limits?

Dashboard related

  • seems you can build a dashboard, but there is no easy way of sharing the URL to your collegues
  • maybe this can be done with the screenboard, but I don't see a reason for a simple URL link share
  • or even better with an embed URL to put it in a webpage or so
  • when I create a view quite often I get the lock of a page, but haven't pressed (on purpose) any lock of the graphs
  • when I set a dashboard to XS in the view, it reset to standard size next time I open it
  • the almost complete (probably trial only) is annoying as it takes quite some space on the top of the dashboard when displayed in presentation mode
  • you can't specify wildcards from something like name:somename* , like in graphite
  • can I create 'public' dashboards with it?
  • allow for custom time range in dashboard display; now you can only select 1h, 4h and have to zoom in manually ; and have to do it every time you open the dashboard

Event stream

Manual event

  • adding a manual event (like a major event happened outside of the usual sources)

Campfire

  • it would be nice to have that chatstream posted to campfire/hipchat etc. directly ; we don't need another chat etc system
  • same for alerts notification

Integrations

Amazon related

  • At integration time , it'd be nice to mention the delay cloudwatch metrics have to reach any dashboard
  • You do remapping of some metrics from /min -> /sec , but the labels it gets f.i. ELB req/min are not adapted , so you wonder if the metrics are correct as they are different from the AWS cloudwatch view
  • You have a way of selecting only certain hosts , but a regex would be nice, also an inverse (don't select these)
  • strange that metrics like ELB unhealthyhosts and ELB 5XX errors or aws.elb.backed_connection_erros are not absolute values but decimal ; eventhough the max is correct. is there a recalculation of the metric there?

Redis

  • I wasn't sure on the integration to install the easy-pip thing, is for python-redis == 2.0 or is >=
  • Even though my agents were already sending redis data, it still had the redis integration to be installed. It could probably detect that automatic

Statsd

  • I was looking for the bind of that daemon, so I grepped for localhost, bind, listen directives. But it seems it's a setting named like make the agent available to outside listeners.
  • It would be nice to bind only a specific daemon

Github

  • I added the integration , but not sure what to do after that. No pullrequest or any metrics are installed

Agent

  • I find it confusing to have the /etc/init.d script start with datadog, the /etc/dd-agent for the config ; a consistent naming would make more sense
  • I did it under ubuntu, and it adds the datadog repo,
  • It would be nice to also have the redis or other integrations requirements in the datadog cookbook?
  • could not find good example for reading logs/parsing as a stream-> metrics

Uploading metrics

  • i'm uploading new metrics to datadog, the http response give status ok , but they don't get into the metric explorer ; is there a delay?
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment