Jeff: We love dashboards too. At Jonoke we had a dashboard that would display information on all our client’s servers, and various ‘Butlers’. This way we were often notified before they called of an issue. Collecting the stats also helps to uncover issues. They take time to write, but they provide an excellent service. As well, all the pretty lights, and graphs moving around give a real sense of accomplishment and knowledge ;-)
If you are using same tool on all the servers I would expect that that slow down would be experienced on all the servers you have it on. From your post that does not seem to be the case. It might be the result of a improper setting on that Windows Server, or by an application on that server. Maybe these can point you to a solution. http://social.technet.microsoft.com/forums/windows/en-US/87e2d994-74c7-4620-96ba-aa6655f18e39/wmiprvseexe-high-cpu-usage-problem-caused-by-bad-startup-program <http://social.technet.microsoft.com/forums/windows/en-US/87e2d994-74c7-4620-96ba-aa6655f18e39/wmiprvseexe-high-cpu-usage-problem-caused-by-bad-startup-program> http://answers.microsoft.com/en-us/windows/forum/windows_7-performance/wmiprvse-using-high-cpu/ea1bb7b7-741e-499b-8bb4-6966ffa2b7dc <http://answers.microsoft.com/en-us/windows/forum/windows_7-performance/wmiprvse-using-high-cpu/ea1bb7b7-741e-499b-8bb4-6966ffa2b7dc> Jody Bevan ARGUS Productions Inc. Developer Argus Productions Inc. <https://www.facebook.com/ArgusProductions/> > On Aug 24, 2017, at 9:05 AM, Jeffrey Kain via 4D_Tech <[email protected]> > wrote: > > I just made a very interesting discovery and thought I'd share it with the > group -- it's timely given the recent blog post by 4D about dashboards. > > We're huge fans of dashboards here at Sweetwater, and our dashboard tool of > choice is Prometheus. We use this on all our major servers -- Exchange, > Postgres, MySQL, and others. We decided to add our 4D Server to the mix. > This server is a 24-core Dell server running Windows Server 2012 Datacenter, > and it hosts, at peak, around 540 simultaneous 4D users and around 3000-3500 > 4D processes. It runs great and it is extremely stable. > > For the 4D Server dashboard, we feed certain statistics directly from 4D via > a SQL function call (poor-man's web service), and we recently installed the > Prometheus wmi_exporter tool which exports operating system-related data from > the Windows Management Instrumentation Service (WmiPrvSE.exe) into Prometheus > format. The wmi_exporter tool is very fast and you have to look fast to see > it consume any resources at all, but when it is running, the Microsoft > WmiPrvSe.exe process periodically wakes up and consumes 4% of our 24-core > server for 10-12 seconds. This happens about once every 15 minutes. > > When this is happening, 4D Server slows to a crawl. Clients start to > beach-ball, pause, or have their network connections broken and throw runtime > errors. Some astute users actually noticed this 15 minute cycle, and after I > dug into it it turns out that is exactly the case and is 100% reproducible. > > We had the dashboard turned on all day Tuesday, and we received dozens of > support tickets from users complaining about slow performance in 4D. On > Wednesday, we ran without the seemingly harmless wmi_exporter service > enabled, and we didn't receive a single support ticket. > > My guess is that wmi_exporter may be asking for a particular statistic that > takes Windows a significant amount of time to collect. We may try to scale it > back and turn our measured statistics on one by one until we can determine > the problem. But until then, we can live without the dashboard (beautiful as > it was on the 60-inch 4K display...). > > Jeff ********************************************************************** 4D Internet Users Group (4D iNUG) FAQ: http://lists.4d.com/faqnug.html Archive: http://lists.4d.com/archives.html Options: http://lists.4d.com/mailman/options/4d_tech Unsub: mailto:[email protected] **********************************************************************
