- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
3 weeks ago
We have a situation where we have ITOM discovery requirements as well as Failover/Load Balancing requirements.
We have 3 Mid Servers each with different user accounts. We plan to have all three servers as part of the same cluster for Failover/Load Balancing. However we also want each server to perform discovery on a specific IP range. Is this a possible use case or will any discovery propagate to whatever available Server?
If not is it possible to have 1 Cluster consisting of all 3 servers and then 3 other clusters of individual servers assigned to their discovery ranges?
Solved! Go to Solution.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
3 weeks ago
Hi @aansah ,
Yes, you can put all 3 MID Servers in one cluster - that part's fine.
But the issue is this once you assign a specific IP range to a specific MID Server, you're forcing that job to always go to that one server. That directly contradicts the whole point of clustering, which is "let any available, healthy server pick up the job." You can't really have both - either the cluster decides who runs the job (true load balancing/failover), or you decide (range-based assignment). Not both at once.
So if Server A is down and it owned that range, the job doesn't fail over to B or C - it just doesn't run, because nothing tells B or C to touch that range. The cluster exists on paper, but for range-locked jobs it behaves like 3 separate MID Servers, not one team.
Your second idea (1 big cluster + 3 single-server clusters per range) doesn't fix this either - a cluster with one server in it isn't really a cluster, it's just extra config for the same limitation.
Fix: if you want actual failover on range-specific discovery, all 3 servers need network access + working credentials for all 3 ranges, so any of them can genuinely pick up the job. If that's not possible (e.g. segmented networks), then just accept these are 3 dedicated servers for their own ranges - clustering won't add failover there, and that's okay as long as everyone knows that going in.
Hope this helps, if it helped you please do mark it as helpful and accept the solution.
Thanks,
Vishnu
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
3 weeks ago
Hi @aansah,
Please refer to the link's below:
https://www.bing.com/ck/a?!&&p=7b58138ce6e0274b05bdf53477e39a636c3127ffaed4dadd5e4de4ad674a48aeJmltd...
Configure a MID Server cluster • Australia ServiceNow AI Platform Capabilities • Docs | ServiceNow
If it is helpful, please mark it as helpful and accept the correct solution by referring to this solution in the future it will be helpful to them.
Thanks & Regards,
Abbas Shaik
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
3 weeks ago
Hi @aansah ,
Yes, you can put all 3 MID Servers in one cluster - that part's fine.
But the issue is this once you assign a specific IP range to a specific MID Server, you're forcing that job to always go to that one server. That directly contradicts the whole point of clustering, which is "let any available, healthy server pick up the job." You can't really have both - either the cluster decides who runs the job (true load balancing/failover), or you decide (range-based assignment). Not both at once.
So if Server A is down and it owned that range, the job doesn't fail over to B or C - it just doesn't run, because nothing tells B or C to touch that range. The cluster exists on paper, but for range-locked jobs it behaves like 3 separate MID Servers, not one team.
Your second idea (1 big cluster + 3 single-server clusters per range) doesn't fix this either - a cluster with one server in it isn't really a cluster, it's just extra config for the same limitation.
Fix: if you want actual failover on range-specific discovery, all 3 servers need network access + working credentials for all 3 ranges, so any of them can genuinely pick up the job. If that's not possible (e.g. segmented networks), then just accept these are 3 dedicated servers for their own ranges - clustering won't add failover there, and that's okay as long as everyone knows that going in.
Hope this helps, if it helped you please do mark it as helpful and accept the solution.
Thanks,
Vishnu
