discovery stuck and ECC queue output in error

[Arnaud9]

Hi everybody,

I'm doing a huge discovery about 6000 CI and it's stuck at 3500 discovered CI and after all the ECC queue in output are in error state instead of processed and the other CI have failed to be discovered. 

Someone has already had this kind of problem ? 

I looked for in lot of KB but I don't find something to help me and maybe me knowledge is limited on certain point 

Other thing when my discovery schedule is small for example less of 1000 CI everything is ok so i thought it's maybe a size problem but where and how can I fix that ? 

[MD AQUIB KHAN]

Hi ,

Please check the Shazzam Batch size. Try to increase the batch size. I guess it is set it less.

Shazzam batch Size

Enter the number of IP addresses that each Shazzam probe can scan. Dividing the IP addresses into batches improves performance by allowing classification for each batch to begin after the batch completes. rather than after all IP addresses have been scanned. The probes run sequentially. For example, the value is set to 1000 and a discovery scans 10,000 IP addresses using a single MID Server. It creates 10 Shazzam probes with each probe scanning 1000 IP addresses. By default, the batch size is 5000. A UI policy enforces a minimum batch size of 256 because batch sizes below 256 IP addresses do not benefit from clustering. The policy converts any value below 256 to a value of zero. The value for this field cannot exceed the value defined in the maximum range size property for the Shazzam probe.

Shazzam Cluster Support

Select the check box to distribute Shazzam processing among multiple MID Servers in a cluster and improve performance. This setting works with the Shazzam batch size. For example, a schedule is created to scan 100,000 IP addresses, with 10 MID Servers assigned to do the work. Each MID Server is assigned to scan 10,000 IP addresses. If the Shazzam batch size is set to 5,000 IP addresses per probe, the schedule runs two Shazzam probes per MID Server (10,000 IP addresses/5,000 per batch). These probes are run in sequence and not concurrently.

[Arnaud9]

Thank you for your answer unfortunately i tried that and it's failing again

[MD AQUIB KHAN]

Hi,

Did you check the discovery log. Also check the ECC Queue. 

try to do quick discovery for one CIs which is failing, I think this will give you some idea. 

[Arnaud9]

Hi, 

Thanks for your answer. I already check the log and everything is ok. I already tried also a quick discovery on a device which has failed and the discovery is succeed so my first idea it's maybe about the capacity of the mid server or something about the network company infrastructure about the bandwidth. Maybe a timeout issue because of that in fact there a lot of flux with the backups, servicenow and others things I don't know. 

Unfortunately I don't have the hand to access to all the infrastructure and I can't do a flow test on the network.

If you have another idea i take it.