We are getting loads of Missing & Partial Responses in CEM, which causes SLA's of 40% to 60%. We know the cause; The TIM has loads of out-of-order bytes caused by packet duplication on the Spans. We have dedup turned on the MTP, but we still get a load of out-of-order bytes. The network team is planning on upgrading the current network infrastucture to account for the increase in SPANS across the network and reduce dupliation. This will take some time.
For now, I would like to disable the Missing & Partial Response Metrics in CEM. Accordin to the GUI, you can't disable them. However, you can ignore them "per transaction". This has been done and the Missing & Partial Response Defects are no longer showing up in the Incident Managent Window. However, the TESS agent in Introscope still reports them, and the SLA's in CEM remain the same, even though the Application shows no Defects in the Incdent Management.
Is there another way to disable tracking of Missing & Partial Responses in CEM? I've heard there may be way to do it by modifying the APM Database using some SQL. It would be nice to allow these to be disabled in the GUI.
Any assistance here for JW?
I have been struggling with these as well for some time, and with multiple sessions with PS & other Techies in CA, havent come across a way to disable these Defects. My major concern is that if we have this set to ignore in DB & we upgrade to a new version of APM, will that cause any data issues?
Thanks for the information. Unfortuantley, I had already set the "IgnoreOnlyRequestErrorComponents" in the TIM settings with no effect on the missing and partial responses. Also, the documentation that is reference, I read before I posted to the community. We are addressing the root cause by looking at upgrading the network equipment to eliminate dedups. One other configuration that I modified was to reduce the Web Filters on the TIM to only necessary IP addresses; reducing the amount of data that the TIM analyzes. However, we still get a lot of out-of-order bytes, which will only be solved by getting rid of the dedups' at the network layer.
Also, the "Ignore" setting in CEM, only works to ignore the incident. The SLA is still calculated on the missing and partial responses. From a customer perspective, this gives the false impression that the application has problems. The application is working fine, but CEM is indicating an SLA of 40%, while Average Response times are well within norms. This could be solved, by having the "Ignore" setting not just stop incident creation, but to ignore the errors as part of the SLA, and to ignore sending it to the TESS agent in Introscope.
Is there anyother way to have CEM (until we get the additional network equipment) disable or truely ignore the Partial & Missing Responses, so that the SLA's are not calculated on Partial & Missing Responses, and the defects do not show in the TESS agent reporting to Introscope.
If this is still relevant, please open a ticket with CSO.