Spooled Notifications are to slow (only 1 notif per sec)

Hello everyone,

we have a problem with the “Notification Spooling” option “Asynchronous local delivery by notification spooler”.
It works as intended and spools the notifications, so that they won’t get lost if an error occurse and so on…
BUT the real problem is the working speed. To create many spooled notifications, it won’t need long, but to send them takes ages, like 1 notification per second, which is clearly not fast enough.

We did a test once with around 800 spooled notifications and the server wasn’t finished even after 20 minutes, so we needed to stop the test there.

Is there any option or possability to increase the working speed of the notification spooler, so it does more than ~1 notif per second?

Sry for grammar mistakes and so on :wink:
Sincerely Kruzgoth

1 Like

Asynchronous delivery should only be used if the delivery method is slow and can have connection errors or longer timeouts.
https://checkmk.com/cms_notifications.html#async
Or if you need to transfer the notifications from a slave to the master for delivery. Then the spooling is the only option.

If you use notifications like mail then it is better to let the MTA handle the spooling.
Other way to reduce the single notifications is to bulk the notifications to bigger “packages”.

If this is not possible then I would first look why I have 800 spooled notifications - there is something completely wrong inside this monitoring system. --> no dependencies set and so on

Nearly one year ago the same question was already on the list [Check_mk (deutsch)] Performance Probleme des Notification spoolers

1 Like

Hi andreas-doehler,

thx for your answer, but that wasn’t my question :wink:
We want to use spooled Notifications because of special reasons i won’t go further in and we want to know, if this specific problem is solvable.

I can only say again notification spooler is not intended for mass processing but for save and reliable transfer. If it should go quick do direct delivery with native spooling if available.

The spooler can replay failed transmission or can wait for re-established connection. This are cases where direct delivery don’t work so good or can block the notification system.

And THAT is exactly the reason we need to know, if the problem with spooled notification is solvable.
We need a stable and reliable transport for our events. :slight_smile:

Can someone help me or has an answer to that?

If you cannot do a direct delivery and you need to deliver 800 messages then you need to wait with the usage of the notification spooler.

But again if I have 800 notifications inside the system I have configured something wrong.

Why is a direct delivery not possible?

Hi,
I got the same problem in a huge environment. A solution for me is using notification bulking.
First I build a notification script that handles bulk notifications. After that I setup a rule for the bulk notification. So you are able to handle hundreds of alarms in one second.
BTW.: It depends on your Alarm receiver. I use an ELK Stack or Netcool Message Bus Probe from IBM which receive JSON over http.

Regards,
Christian

Hi ChristianM,

that sounds good, but we need these alarms asap and can’t wait for the notification bulking to collect hundreds of notifications first or wait a specific time.
It looks like there is no solution here from Check_MK, which is kinda sad.

Thanks

Sincerely Kruzgoth

The solution from @ChristianM is instant delivery. Bulk time set to 5 or 10 seconds and that’s it. I don’t know any notification system what can handle some hundreds of message in a shorter time without any special queueing system.
Your system was this not Netcool?

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.