Failing Web App Availability monitors

Dashboard Server
,
1

I’ve got several web monitors that have a large amount of ‘flappage’. The error code I get is 80090321. A google search seems to indicate:

‘SEC_E_BUFFER_TOO_SMALL 0x80090321 The buffers supplied to a function was too small’

Anyone else experiencing this and got a fix? I’ve read something about WinHTTP buffer errors?

1 Like
2

Hey Peter,

How many web checks are you running on that server? If you use the built in web availability monitors from SCOM the agent’s resources gets congested relatively quickly.

You can check this by filtering for EventID 2023 and 2120 in Operations Manager Log:

The health service has removed some items from the send queue for management group since it exceeded the maximum allowed size of 15 megabytes

If that is the case there are 3 things you can do:

  • Change the parameters of the SCOM agent running the health check: https://blogs.technet.microsoft.com/kevinholman/2014/06/25/tweaking-scom-2012-management-servers-for-large-environments/
  • Run the webcheck on another server (if you are not using the resource pool)
  • Take a look at the URLGenie MP, which is a lot more scalable. https://gallery.technet.microsoft.com/scriptcenter/URLGenie-Management-Pack-771bf58c
Br, Jasper
3 Likes
3

Hey Peter,

What happens when you open the URL manually? Does it work?
I also have a solution on my blog to monitor websites from an external perspective here. The only thing that it is not great at is response time monitoring, as it seems the server performing the checks are mostly from one or two datacenters located in the US.

Keep in mind that GSM is limited to 25 checks, and each datacenter you enable monitoring from essentially consumes 1 check.
Br,
Jasper

2 Likes
4

Thanks for the suggestions Jasper, There are 68 web app availability tests and 12 web transaction tests run from a resource pool of two management servers. These management servers are also doing *nix monitoring (70) and network devices (75). That doesn’t seem too much load to me - or is it? The servers don’t look to be particularly taxed.

I don’t have those EventIDs in the Operations Manager log on the two management servers. The management servers, and the bulk of what they are monitoring are all virtual servers on VMWare. I’ve started looking at Global Service monitor to do some outside-in monitoring.

I did look at URLGenie briefly before, time I took a closer look I think!

5

Hi Jasper, 99% of the time I get the 80090321 error code the website is fine. Thanks for the link, I’d already planned to look at this - I think I’ll bump it up my list and try and move away from internal checks as much as I can.

For GSM, I plan to just use it for our most important websites from Dublin and Amsterdam, that should give me 12 sites and 1 test spare!

Thanks for the help!