Question - Shutdown stats #EOC and #KERNELVERS for almost all of my Broker kernels from this past IPL have dramatically increased for reasons unknown to me. All the other HWMs and COMMAND #'s look about within a normal range for the period of time (2 weeks) but these two stats are WAY out there.
As a matter if fact, one of the Brokers historically shows 0 EOCs every cycle and now this past cycle shows 12620 EOCs.
Any ideas on what I should be looking for if it occurs again? I have no idea about code changes on either side in the applications, but the fact that they all changed like this regardless of test or production makes me think it isn’t necessarily caused by an application making more than the typical number of calls.
Hi Rolf,
Yes, that’s possible. There’s only one other person who might be running SMH that my group would be aware of. He’s getting ready to migrate the XMLRPC servers from Windows to Unix - so he possibly was running SMH but he doesn’t usually.
I’ll check with him tomorrow to verify he left it up and running or something.
My group usually goes in, looks at what we need to see and then logs out so the monitoring tool is never really “up” all the time.
I’m going to ask the obvious dumb question — could an increase that big be caused by 1 person starting SMH and staying on?
Mogens - I would activate a trace but it is appearing on all Broker kernels - not just one. And I don’t know when it started or if it is still occurring---- so I’m going to hold off on that for now (knowing how much output gets generated).
Rolf - We have one person who said he has been running SMH continously for the last week or so…but he doesn’t see how that would be impacting all the other Brokers stats because he only defined one of them in his instance. He also mentioned that someone else was using Broker to “load” some type of data to test with - but I would expect to see increases in SEND and RECEIVE commands if that was the cause.
I’m not going to worry too much about it unless I see it again next cycle. However, if you have any other suggestions or things to look for, I’ll take any advice.
what version are you on? you didn’t upgrade recently? have any of the applications upgraded to newer versions?
I ask as recent versions of the server and client tools issue KERNELVERS commands to verify version and security settings automatically. So if you upgraded your XMLRPCSERVERs, for example, they would issue lots of KERNELVERS.
You might request a performance and tuning review (ok, ok, so I’m plugging GCS services!).
That might explain part of it… I’ve recently upgraded my SNH to v9.5.1 and someone else has recently gone to v9 - not sure when.
The XMLRPCSERVERs are something I’m not in charge of - but I think the guy who works with them may have upgraded to v9 recently. Our XMLRPCSERVERs only run on certain kernels though, not all of them so I’m still not sure why I saw increases on the other kernels.
I’ll keep that in mind (the upgrade stuff) and also consider the performance and tuning for a future date. We had one about 2 years ago…but currently are in the process of doing upgrades to a bunch of SAG products and will probably put the P&T off until the dust dies down from that.