Hello All,
Suddenly all our CRM agents are complaining a significant delay when opening a record from different entities (incident, Contacts ...etc). Views are working fine on the same time the slowness happens. We have our CRM deployed on AZURE cluster.
Any ideas on how to start troubleshooting this issue.
*This post is locked for comments
Hi AWT,
I see you are using 2 FE Servers. Is the deployment manager configured for an NLB configuration? - See following blog: blogs.msdn.microsoft.com/.../network-load-balancing-microsoft-dynamics-crm-2013
Additionally, how is the affinity set against the NLB? If it set to single (Multiple requests from the same client must access the same member; useful for clusters within an intranet.)? Other NLBs might use different naming convention for that. However the functionality is the one I've described. Additional Information can be found here: technet.microsoft.com/.../bb687542.aspx
If you try to bypass the NLB, for example edit the HOST file on one of the clients, to connect directly to a CRM FE do you see any difference ? Try this against both of the FE.
Another aspect - you are on CRM 2015 RTM - did you consider upgrading to the latest UR 7.0.3? I wouldn't put much hope in this solving the issue, however it's always best to be on the latest UR, to exclude any potential bugs that can be in between builds.
Thank you very much for your response.
This is the result of running performance center, as you can see the Read-Ready stage is taking too much time.
I want to highlight the following as it could help:
Hi AWT,
What CRM build are you on? Were there any updates installed recently or any security changes on the deployment?
Usually performance issues need to be tackled from multiple angles.
My suggestion is to define a clear scope to know what you are after. Select one scenario - for example, focus on the slowness when opening a Contact record.
Some suggestions to try when facing performance issues
Please keep in mind, that when troubleshooting performance issues, tracing can also impact - is it does I/O Operations and additional processing. So you would need to be aware of this aspect when enabling tracing. Best is not to enable all at the same time, but rather try to narrow down step by step, until you have a potential suspect on where the issue might be coming from. Afterwards you can enable against that component more detailed tracing.
Hope this helps,
Radu
Mohamed Amine Mahmoudi
83
Super User 2025 Season 1
Community Member
52
Victor Onyebuchi
6