by Jamie
I would really like to have an option for configuring dependencies for
monitors. Consider this scenario: Lets imagine 50 (or whatever number of
web applications) running at a given data center or server. If the
network or server goes down, I would receive alerts/notifications for
all of those web sites. It would be better if we can configure a network
map (such as in Icinga) or on a simpler level, have an ability to define
monitor dependencies. In my scenario described, those 50 web site
monitors would be dependent upon the gateway device or server upon which
those 50 sites are dependent.
Without this feature, there is potential with a catastrophic type
of failure to 1) burn through all notification credits very quickly 2)
overload on call folks with an unmanageable deluge of SMS and phone calls.
In my example, those 50 web sites might be dependent upon the
server being reachable (ping), and that server might be dependent upon
a gateway device (or switch). I am thinking this is going to be a
pretty important feature for anyone monitoring more than a couple
services at a single location.
Like (4)
Reply
Replies (33)
The request has been forwarded to development team to check the
feasibility of having this feature implemented and add it in our product
roadmap.
Thanks.
Rafee
Like (0)
Reply
by Gibu Mathew
We acknowledge the importance of this feature.
We currently do have the ability to monitor the status of gateway
servers and other hardware present inside the data center via the
Site24x7 On-Premise Poller. I guess we just need to give a way to
"associate the status".
BTW I would like to hear your comments on whether you would like
to see the "dependent monitor" associated on a per monitor
basis (i.e every website) or at a Monitor Group basis. Associating
at a Monitor Group Basis makes it easier to manage when you have
hundreds of monitors.
Like (0)
Reply
I think associating at a monitor group basis or individual monitor basis
both make sense. Having the ability for associating groups certainly
makes sense in my use case.
Jamie
Like (0)
Reply
by IT Marcaria
I would take this association a few steps further.
If you have a server monitor and then all kind of monitors related,
i would love to get a full report view and to include data from all
monitors in case of an outage.
For example. Lets say i have an IIS server that works with an
MSSQL database. I would have server monitoring, IIS monitoring and
MSSQL monitoring. Additionally i might have Website and web page analyzer.
If there is an outage or an incident related to performance.
Today i have to manually gather data from all monitors to see whats
happening. While if this monitors are correlated, you can send a
full view of all data or get any parameter that's behaving out
of the average.
Another feature that could empower the above idea, is to Sync the
poll of related monitors. That way, the information gathered from
different sources would be from the same time frame.
Like (0)
Reply
by Gibu Mathew
I agree it is important to see monitors that are closely related in
a more intuitive manner. Currently there is no automatic modelling of
monitors on the same physical host. Hence this has not been done. This
is an important enhancement we have in mind. This will allow us to
atleast list the monitors and its status in the Monitor Details view.
Probably inline in the "Summary View".
>>
Today
i have to manually gather data from all monitors to see whats happening.
That
said, we do have plans to provide a customizable dashboard. But that
is some time away as we are working on getting our new client done.
We have the Monitor Group Concept for a bit more advanced use case.
Say you have a Web App that uses 2 DBs, 10 App Servers, 10 physical
servers. In order to see the status of these infrastructure apps in
one place when there is an outage or for general reporting needs we
have to group these "Monitors" under a "Monitor
Group". The Monitor Groups can ideally serve the purpose of
representing status of a "Business Application" in this
context. This is something we will see more in the new reports that we
are working on. The new reports would come with the new web client we
are working on.
>>Without
this feature, there is potential with a catastrophic type of failure
to 1) burn through all notification credits very
>>quickly 2) overload on
call folks with an unmanageable deluge of SMS and phone calls.
I re-read your original post and
found that we should have an extra protection for this overload of
alerts. For voice calls, we do have some safeguards in place.
After a few calls we will begin to "consolidate and make a
call". This is not present for SMS and EMail Alerts. However
I guess we could look at providing the following new features:
1) A web link in the
"SMS" or EMail to
"Turn
Off Alerts for the next hour" for that "Contact".
2)
Option in the Web Client > Operations Tab to "Turn Off
Alerts for all users for the next hour" at an account level.
Thank
you for your comments.
Like (0)
Reply
I did experience that the phone call alerts do start to consolidate, the
safeguard you mentioned, where multiple alerts are delivered in one
phone call. So that is very good. I think the same thing makes sense for SMS.
Also, and ability to acknowledge problems would also go a long way,
because I don't really like the idea of suspending monitors
because then we don't get real uptime/downtime data, and there is
risk of forgetting to un-suspend. Second, just disabling alerts per
user has the same risk of forgetting to re-enable alerts and getting
configuration back to where it should be after a problem. I think your
options 1 and 2 go a long way towards addressing that issue, however I
also think the real need is an ability to acknowledge problems in bulk.
Hope that helps!
Jamie
Like (0)
Reply
by Gibu Mathew
The ability to acknowledge problems in bulk will come in to the product
later this year. We hope to include it in the product by means of
introducing an Alarms Tab. I can share some screenshots of a prototype
that we have internally.
If you are interested in looking at it, please send an email to
"support@site24x7.com Attn : Alarms Tab Prototype".
Thanks
Gibu
Like (0)
Reply
by syed.hussain
The dependency association should also be present for agent based server
monitors. For example, I have my firewall monitored externally, it went
down, basically taking the complete internet down, all the servers
behind that went offline in monitoring dashboard, the system sent
multiple alerts and SMS, although the servers were up and running the
only thing was internet went down. So it would make sense if those
monitors could depend on another monitor so that false alarms do not
come up.
Like (0)
Reply
by Gibu Mathew
Thank you Syed for this use case.
@Sabarinathan, if you have any comments on how we handle and
prevent lots of alerts getting generated for server monitors in the
case when the firewall or some edge device at the customer end has a
problem use case, please post here.
Like (0)
Reply
Thanks for bringing up the Feature request. Currently we trigger alerts for all the Server Agents and it dependency monitors. We will take this feature immediately and will let you know once updated as soon as possible.
Like (0)
Reply
by Gibu Mathew
Sorry to say we have not started on this yet. However I like to
reiterate we understand the importance of this capability.
Like (0)
Reply
by Jamie
Well it happened today. A network outage not overcome by multiple
redundancies caused hundreds of alerts, sms, and calls to go out to
multiple personnel. This continues to be a hot item for our use case.
If only I could have each of my host groups have a parent
dependency of the cluster IP (ping monitor) this problem would be
alleviated. Glad to see that this is under review, and just wanted to
give it a friendly *bump*.
The good news? We definitely knew there was a problem thanks to
Site 24x7.
Like (0)
Reply
I also added Sub Grouping as a request which would play well with what
you are looking for. Thanks for bringing this up.
https://forums.site24x7.com/topic/sub-grouping
https://forums.site24x7.com/topic/sub-grouping
Like (0)
Reply
Jamie, as for SMS here is what we do to limit the crazy SMS usage.. We
only have Site24x7 SMS alert the highly critical people and checks and
we use PagerDuty for all other alerts and Site24x7 is integrated with them.
I do understand the crazy issues if a switch goes down and it sets off a chain reaction for everything under it.
https://www.pagerduty.com/pricing/
I do understand the crazy issues if a switch goes down and it sets off a chain reaction for everything under it.
https://www.pagerduty.com/pricing/
Like (0)
Reply
Thanks framirez. I had considered pagerduty when I originally set about
to have a workaround. Thanks for confirming the integration works for
you. I was hoping to avoid subscribing to yet another service, but it
may be my only option right now.
Like (0)
Reply
by IT Marcaria
Perhaps a faster way to see this implemented would be to have the
feature of monitors in multiple groups and also the feature of adding
sub groups.
After that is necessary to include the logic of checking if a group
is already alerted or not before sending an alarm.
Like (0)
Reply
Hi Jamie,
Thanks for your patience. We are starting to work on this feature
now. Will keep you updated on possible availability date once we get
to that point.
Like (0)
Reply
Hello,
Sorry the feature was delayed due to other priorities earlier.
We have now started working on this now. We will have it
released in 6 - 8 weeks time.
Will update this post once it is live.
Thanks for your patience on this.
Raghavan
Like (0)
Reply
Hi Kamil,
Thanks for updating on your requirement. Just wanted to update on what is planned on this feature:
- Support for allowing monitors under multiple groups
- Support for creating sub groups under a monitor group up to 5 level.
- Support for setting up dependency monitor / group for a Monitor or a Monitor Group.
- Support for configuring threshold for group failure.
- Suppressing alerts for a monitor when dependency monitor / group is Down.
Let me know if you have any further requirement on this feature.
Regards,
Raghavan
Like (0)
Reply
by kkeski
Raghavan,
I think this list looks good. I'd just like some clarification on alerts. Will group based alerting be coming for the ride with this change (associating a monitor group with a user group as that's where alerting is specified)? Currently it's only possible to specify an alert policy at the monitor level. I think having a group level alerting profile would be useful and perhaps even required with this new functionality.
Much appreciated,
K
Like (0)
Reply
Going through old forums as a new customer. Monitoring Group alerts still isn't implemented and is something we're in need of. I created a new forum/idea specifically for this https://forums.site24x7.com/topic/alert-on-group Please upvote so it gains visibility! :)
Like (0)
Reply
What's the ETA on releasing the dependency feature? For us
this is the requirement before we deploy this to our entire production
environment. Thanks for the info!
Best,
Julian
Best,
Julian
Like (0)
Reply
Hello All,
The much awaited alert suppression feature is now live.
Users can now go ahead and set up a dependency monitor. It is mandatory to associate the monitor to a monitor group and then map it to a dependency monitor. You should also enable the suppress alert option in the tab. Please take a look at the screenshot given below for more clarity.
Regarding the other enhancements as listed above by Raghavan, we are working on the same and shall update this thread once done.
Cheers!
Sushma.
The much awaited alert suppression feature is now live.
Users can now go ahead and set up a dependency monitor. It is mandatory to associate the monitor to a monitor group and then map it to a dependency monitor. You should also enable the suppress alert option in the tab. Please take a look at the screenshot given below for more clarity.
Regarding the other enhancements as listed above by Raghavan, we are working on the same and shall update this thread once done.
Cheers!
Sushma.
Like (0)
Reply
by Brad
Hi
Love the dependency monitor feature, but had occasion this weekend to see the need for a possible change to it (or maybe this exists already?):
I needed to reboot a VM host running many VMs, all of which have the host as their dependency. I put the host's monitor into 'maintenance' on 24x7 thinking the individual VM monitors would take this into account and not alert me as I shut them down cleanly. No such luck.
What I am guessing is the dependency monitor is only looking for a DOWN condition on the target.
Could you add 'maintenance' into that consideration? Since I can see how this might not be wanted as a ubiquitous feature, can you perhaps add a toggle on the maintenance scheduler...something like "let dependents know"...or something so that I don't have to put all of the dependent monitors into maintenance mode individually?
TIA
-Brad
Like (0)
Reply
The ability to select more than one dependent monitor would be awesome.
For example:
we have a page ( https://www.site.com/page1 ) that is dependent on ( https://www.site.com ) to be up but it's also dependent on a server ( server001 ) to also have port 443 open.
Or if we are working on site ( https://www.site.com ) then ( https://www.site.com/page1 ) and ( server001 ) should not send alerts because we already know the site is being worked on.
For example:
we have a page ( https://www.site.com/page1 ) that is dependent on ( https://www.site.com ) to be up but it's also dependent on a server ( server001 ) to also have port 443 open.
Or if we are working on site ( https://www.site.com ) then ( https://www.site.com/page1 ) and ( server001 ) should not send alerts because we already know the site is being worked on.
Like (0)
Reply
Hi,
Thanks for raising the request with your usecase.
Do you think it is good to support configuring Monitor Group as dependednt for a monitor, thus it allows any number of dependent to be configured ?
Raghavan
Like (0)
Reply
Hi Brad,
Alert suppression will be done for dependency monitor maintenance state as well. Apart from that, in the case of VMs, even without configuring explicit dependency, if the host monitor goes to maintenance state, VM monitors will also go to maintenance state.
Looks like you are facing some issue, please send a mail to support at site24x7 dot com for analyzing the issue further.
Thillai.
Like (0)
Reply