Comware

In PCM 1.5, when events were acknowledged, the Dashboard reflected it immediately. In 1.6, it does not. I don't know where or how the refresh is set or done. I cannot trust the Dashboard to give me current status and must squander another monitor just to view my traplog. I already have three monitors dedicated to PCM alone and my office is starting to look like Mission Control at NASA.

This is not good. What can be done about this?

What? Nobody else with this problem? Nobody to comiserate with?

Slow to refresh is an understatement. It has more of a California attitude... updating if/when it feels like it.

I should mention that 'About' reports as B.01.60.A, that I have both the Find-Node-Patch-PCM-1-6 and the pcm-1-6-a-patch installed.

1.5 ran decent on our dual 550mhz 1gb server...
1.6 killed it...so until we get new hardware, no pcm+ for us ;)

Anybody else having this issue?

Anybody care?

I thought I would clean the slate and delete everything out of my traplog just in case there were some ghosts in there from my 1.5 days but still, if I get a single alert, the Dashboard shows bogus numbers of various alerts, not just the one that is in the traplog. I tried rebooting but that made no difference.

I opened a ticket with nethelp and was initially told that that was how it works, but on my insistence, they will dig deeper.

Anyone else getting bogus numbers?

OK, I have a pet theory on this. In the Preferences, under Events, there is the Ignore List where I had all three checked. My theory is that when one of these alerts come in to PCM, it bumps the counter even though it is ignored. Once I have something in the traplog for the severity that is showing bogus values, if I delete just one of them, the number will for a short time be correct. If I clear the log and uncheck everything in the Ignore List, the numbers are no longer bogus.

<UPDATE>Well... so much for my theory. :( I was just about to hit the Submit button on this post when suddenly I got bogus numbers again. ARGHH!</UPDATE>

That still does not explain the fact that acknowledging the traps does not clear them from the Dashboard. In 1.5, acknowledged events were not shown on the Dashboard.

A problem that has be investigated and acknowledged should be so represented on the Dashboard. If HP wnats to display acknowledged events as well, then the colourless high water mark that the Traffic Monitor should be used. As it is, this Dashboard is useless for monitoring an overview of traplogs.

It is unreasonable to have to delete the traplog entries to clear the Dashboard. It is common practice to investigate and acknowledge an alert but not delete it because it has not been resolved.

Does anybody else agree with me on that point?

Another update... NetHelp is escalating my case #3210163101 up to Division.

BTW, In respose to Preston, I don't know if PCM+ will use both of your processors or not. I had tried running on a dual 450 server but when I upgraded to 1.5, I moved it to a simple desktop machine running XP. I am running PCM+ 1.6 on a single 2.8 ghz P4 CPU with 2 gig of RAM. The Commit Charge has never reached the 1 gig mark but the CPU does get a bit of a workout at times.

I did find a bit of a workaround. After I acknowledge the events, if I delete any one event, the Dashboard will update. Get another event and some bogus ones are likely to show up again, but if I sacrifice an event, the bogus column will go away... as long as I keep a few events around to sacrifice to the PCM Gods.

I had hoped that this would be fixed in version 2.0 but it seems to have gone from bad to worse. Now with 2.0 Gold, the dashboard never resets on acknowledged and the workaround no longer works. :(

It turns out that someone at HP decided that overall totals rather than new (unacknowledged) events should show in the dashboard summary. The last word was that an enhancement request will be submitted to have the overall totals as a ghost outline and the "new" events in colour like the Traffic Monitor does now.

Since the overall totals will drive the auto-scaling of the bar chart, there is the potential that the "new" bars will get dwarfed si I hope that will be taken into consideration.

I cannot for my bare life see why overall totals are relevant on a dashboard like that. The dashboard should give up-to-date info on "what's going on right now?", and should not confuse users by including old news.

Cheers,
Anders :)

Anders,
I fully agree. The acknowledged totals mean nothing to me. The only thing I want to see is the unacknowledged totals.

I generally acknowledge events as soon as I see them and only delete them after I resolve the issue. I like to keep some of them not because I haven't resolved the issue, but rather as a reference to go back to later.

On further reflection, because of the auto-scaling, it will tend to diminish the at-a-glance impact of the summary, so I am starting to regret the compromise suggestion of the ghost highwater marks. I am now thinking that highwater marks should be settable as a preference and with them off, only unacknowledged would show.