wvitalec replied on Thu, Sep 24 2009 2:41 PM

rated by 0 users

I can, but it is slow, like any web console would be when changing multiple elements.

I will just deal with it as is and figure out my polling behind problem.  It really seems that this latest version is not an improvement over V9.1.

MNPS_G replied on Fri, Sep 25 2009 9:16 AM

rated by 0 users

I have been feeling like somewhere along the way in Ver 9.5 that my polling engine has been behind, but haven't really seen anybody else mentioning it until this thread.  I am up to SP4, and somewhere along the way nodes were constantly going up/down in Orion, but if I polled them myself they were fine.  Stopping Netflow collection has stopped the up/down, so I assumed it was DB related problems, but the following is making me wonder. 

The other interesting item is that the Polls Per Second Tuning recommended settings change somewhere along the way for 'Maximum Node and Interface Status Polls'.  It had been around 130 per second, and when I went in there to try and tweak things to help stop the up/down events, it is now recommending 88.  That is with no hardware changes to the machine and the only software changes have been the SPs for Orion. 

I'll open a ticket for this, but just wanted to chime in as well. 

wvitalec replied on Fri, Sep 25 2009 9:45 AM

rated by 0 users

I definately think there is something in the V9.5 database configuration/indexing that is causing this.  I have also noticed, today, that I have numerous interfaces that are polling unknown.  The problem with the web console is, if you try to rediscover them you get the same status; however, if you do this in the System Manager you have the ability to add them as up and remove the 2 unknowns.  This web console is really a huge limitation.  I strongly request the System Manager be brought back at least until all these bugs with the web console can be worked out and tested completely.  I can't afford to be a test subject for all these bugs in the web console in an extremely large production environment.

I check my polls per second every day and they are always what is recommended so I know the polling up and down as you state is not from that.  I had the polling almost perfect with V9.1.  I never saw the polling go over 5% behind for any polling engine.  Now I am seeing it jump up to as much as 30% behind throughout the day consistently.  The issue I have with this is NO changes were made to any devices or my network.  The ONLY change has been V9.5 and the numerous service packs.  I really don't like sounding like a broken record, but I have to considering the environment I am monitoring and the cost to the business if I can't properly do that.

denny.lecompte replied on Mon, Sep 28 2009 10:38 AM

rated by 0 users

wvitalec:

I definately think there is something in the V9.5 database configuration/indexing that is causing this.  I have also noticed, today, that I have numerous interfaces that are polling unknown.  The problem with the web console is, if you try to rediscover them you get the same status; however, if you do this in the System Manager you have the ability to add them as up and remove the 2 unknowns.  This web console is really a huge limitation.  I strongly request the System Manager be brought back at least until all these bugs with the web console can be worked out and tested completely.  I can't afford to be a test subject for all these bugs in the web console in an extremely large production environment.

I check my polls per second every day and they are always what is recommended so I know the polling up and down as you state is not from that.  I had the polling almost perfect with V9.1.  I never saw the polling go over 5% behind for any polling engine.  Now I am seeing it jump up to as much as 30% behind throughout the day consistently.  The issue I have with this is NO changes were made to any devices or my network.  The ONLY change has been V9.5 and the numerous service packs.  I really don't like sounding like a broken record, but I have to considering the environment I am monitoring and the cost to the business if I can't properly do that.

Just for the benefit of other folks reading this thread, one of our Devs is on a web meeting with wvitalec.  There are at least some circumstances that seem specific to his installation  that are causing his performance issues.

wvitalec replied on Wed, Oct 7 2009 8:47 AM

rated by 0 users

Why did the Disable Session Timeout stop working in SP4?  It is such a pain to have to constantly log in when automatically logged out.  This has even happened in the middle of adding a node and interfaces.  When is the web console going to be able to manage a large amount of elements?  I can only edit properties on 2 or 3 interfaces at a time.  If I try to edit 10 or more for example, the console never refreshes.  Bring back System Manager.  This web console does not handle enterprise size environments.

denny.lecompte replied on Thu, Oct 8 2009 12:52 PM

rated by 0 users

wvitalec:

Why did the Disable Session Timeout stop working in SP4? 

We haven't heard this from anyone else.  As for web console performance, you seem to be experiencing problems that other users are not having.  We have other customers with enterprise environments who are using the web console quite successfully, especially with 9.5 SP4.  Your request regarding System Manager has been heard and recorded.