I am also encounter the situation to disable group. I want to store the alert one by one into my database when it happens. As grouping alerts, it would duplicate alerts in the group when one alert was resolved. 在2017年10月13日星期五 UTC+8 下午4:19:22<Brian Brazil> 写道：
> > > On 13 Oct 2017 09:04, "Rob N ★" <[email protected]> wrote: > > On Fri, 13 Oct 2017, at 06:29 PM, Brian Brazil wrote: > > Sending notifications into slack/email like this is what I'd consider to > not be very useful, as the important thing about a alert/notification is > that a human needs to look at it. What you want here is usually a ticketing > system, so these alerts can be assigned to someone to process at. > > > That's more about processes and scale. I have three operations people. I > have no more than two or three of this warnings on any given day. > Eyeballing Slack and reading our email is more than enough to keep track of > it. > > If they're just "for information" I wouldn't even send them to the > alertmanager. > > > Slight tangent, but does anything else implement the alertmanager API? > > > I'm only aware of company internal things. > > > Or phrased another way, what other methods exist for getting this kind of > stuff out of Prometheus? Just make a dashboard/console that runs the > interesting queries? > > > Using PromQL is one option. > > > > For the more ticket type things the questions are more what is the batch > size the human works with. For example if I'm processing them once a day > I'd like them all in one notification so I can handle them in batch, as > they likely are related problems with a single common cause. If each > requires substantial unique engineering effort to deal with on varying > timescales, then I might want one ticket each - and to look at redesigning > the system not to produce so much operational work. > > > The receiver gets full details of every alert in the group, if you want to > show more than the defaults (we only put a summary in things like Slack to > avoid spamming your channels with 10-page notifications every 5 minutes, > whereas email has everything) then that can be configured with notification > templating. > > > Yeah, I think we're definitely talking about difference in scale. In my > case I just don't have enough warnings to be overwhelming, but do have > enough that I want to know about. > > Also you say you want related things in one batch, but my point here is > that these things look related, but aren't. > > > This is what the power of routing and groupings is designed for, so you > can configure business rules like these. > > > Don't get me wrong; I totally see why the grouping behaviour is useful if > that's what you need, but in this case its in my way. > > Anyway, its not a showstopper, and I have a few other ideas for how to do > what I need, so I'll go and give those a try. I do appreciate the > discussion; it's helped to clarify a few things in my mind. Cheers! > > Rob N. > > > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/1d9e1dc8-f6e6-4804-b5dd-b451c9f3ded7n%40googlegroups.com.
