Remote Hypertrend Traces Incomplete and Sporadic V6.02

[erblock]

Related to the thread on Hypertrend traces through simlinks from remote subnets.  However, this thread details problems with Hypertrend traces run in clients on the same subnet as the computer where the server process and citadel database are resident.

There are four attachments (PDF) showing shots of two different time slot views.  The views are all taken from an identical client process.  Two from the process running on the same computer as the server process and citadel.  Two of the shots from same client process running on a different computer, but on the same subnet.

All the clients were run in Lookout 6.02 Full Development version.

These are sample results.  I have confirmed on two other clients and via the internet that this behavior is sporadic and varies by machine running the client.  I have also noticed that on a different day, the same computer that was missing (display) data on the day before, will show the data correctly at a later time.  Things seem very sporadic and don't seem to have any specific pattern.

First three attachments are with this initial post and fourth will follow immediately after.

Please help and let me know if I should start an offical support request on this.

Thanks,

Ed

[erblock]

Fourth screen print here....

[erblock]

Some more shots of Hypertrend irregularity.  All identical client process files.  Client-Local is running within 6.02 full development version.  All remote clients are running web-client with clean download of same server computer exported web client.

Notes - filename definitions:

Client-Local: Client is running on same computer as server process and Citadel.

same subnet: Client is running on computer within same subnet as server process and Citadel computer

local subnet: Client is running on computer that is off same switch room but is on a different class C address space (local DHCP address space).

Comments anyone....please!

Cheers,  Ed

[Ryan.S]

Have you installed the 6.0.2 updates? I mean the 1.4MB one.

I notice that there is a fix on the hypertrend in the updates.

Message Edited by Support on 02-04-2009 04:44 PM

[erblock]

Well,

The server computer (running server processes and Citadel) has the latest updates.

The web clients (second set of views in the .zip file) of course, do not have any updates, since they load from NI's web site and the web client has not been kept up-to-date....

The clients on the first set of views (views 1 and 2) all had latest updates (full development versions of Lookout).

New data today....BREAKING NEWS:  My first seemingly totally working hypertrend set was run on a web client machine on the local subnet.  The bad news is....I have not been able to reproduce that level of sucess on any of many other local clients (5 separate machines on same subnet to-date have at least major Hypertrend display irregularities).  The sucessful client used default web client download/install from NI site.  It is significant to note that the same machine was cleaned of a full 6.02 development version with latest updates (full uninstall) prior to clean web client install.  Identical client process had major Hypertrend irregularities when run in previous full development version.  I have since tried repeating the process on a machine that is on a remote subnet, but the symptoms did not change.

[erblock]

It seems related problems in this thread have been ignored for many months now.

http://forums.ni.com/ni/board/message?board.id=170&message.id=166278&query.id=499010#M166278

So whats the deal here guys and gals.  You going to fix or provide reasonable workaround solution for this or not?  I mean afterall, the issue affects your flagship product users also, not just your left behind Lookout users...

[Ryan.S]

I have talked with the developer. This is an expected behaviour sometimes.

In the first screenshot, the data hasn't been replicated to local cache in time, so hypertrend doesn't show it. After a while, the replication is done and the data on hypertrend comes out completely.

This is because either too much data needs to be replicated each time or the trend width is too big.

You can try to decrease your trend width to see what will happen.

[erblock]

Quote: I have talked with the developer. This is an expected behaviour sometimes.

______________

Well,  It doesn't happen just "sometimes", and I certainly believe that from a customer standpoint, this certainly should NOT be "expected behavior".

It only takes a few seconds for an entire screen update for those 72hour, complex data trends.  So I don't see how replication has anything to do with it.  I can take the missing data area and zoom into only an hour or so and the data still does not refresh.  Additionally, you must remember...this BUG is only present in Hypertrends running on remote machines.  If the same client is run on the local machine, the data will always fill in empty areas quickly.

So, relative to the remote client process, is there a way to adjust the cache replication time constants?

Why does this problem not exist on local machine clients?

I'd really like to make sense of this, and I'd also like to see a fix SOON.

Ed

[Ryan.S]

good news, we have fixed a similar problem in DSC 8.2. I mean the updated Citadel with DSC8.2 might solve your problem. Let me figure out what you need first and find a way to give you the fix.

[Ryan.S]

Ed,

could you provide a personal email address so that I could contact you directly and might give you something to test.