It is currently Tue Jun 27, 2017 3:39 am


All times are UTC




Post new topic Reply to topic  [ 29 posts ]  Go to page Previous  1, 2, 3  Next
Author Message
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Tue Jul 19, 2011 5:49 am 
Thanks for the writeup RN, very much appreciated.

CJ, a few thing you might check (if you haven't already solved this problem):
1. After you apply the "*" idle-pc value, go to Edit > IOS images and hypervisors.
Highlight the correct image and ensure that value is entered in IDLE PC field




Top
  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Fri Aug 12, 2011 3:16 am 
Offline

Joined: Fri Aug 12, 2011 3:12 am
Posts: 2
Hope we could up this thread...am learning new things a lot here...


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Sat Sep 03, 2011 8:00 pm 
Offline

Joined: Sat Sep 03, 2011 7:53 pm
Posts: 2
Hey guys. I know this is a rather old thread that has been revived, but I'm having some trouble calculating an idlepc value also.

I've used GNS3 many times and not had issues with this. I am now attempting to use this on my little EEE PC that is probably quite underpowered for this task, but it's the only PC I have with me right now.

I have tried all of the suggestions, but I can't get an idlepc value with a *. I've tried over and over, and I've tried selecting other non* values to see if it would throw a *'d value afterwards. No dice. My CPU utilization jumps around a bunch, somewhere from 50-80% typically, regardless of the value used.

Any pointers would be great! BTW, I'm using the IPBaseK9-M 12.4(10a) IOS for the 2621. The image is unpacked. I can switch to another image if need be - that's the only thing I haven't tried.

TravisT


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Mon Sep 05, 2011 12:30 am 
Offline

Joined: Fri Mar 05, 2010 11:33 am
Posts: 1494
Location: Australia
How "underpowered" is your EEE PC? Is it able to run VMWare? You may actually get better results running GNS3 in a linux VM. (like GNS3 Workbench)
Alternatively, do your IDLE-PC calculation on another PC that does have some extra power, then use the same value on the EEE. As far as I can make out, once you have a good idle-pc value for a particular IOS it is good across all host operating systems. (Discussion welcome)

_________________
RedNectar
http://rednectar.net
@rednectarchris
GNS3 WorkBench-a VMware image of Ubuntu with GNS3 and VPCS installed and a collection of exercises/labs


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Thu Sep 08, 2011 2:21 pm 
Offline

Joined: Sat Sep 03, 2011 7:53 pm
Posts: 2
I didn't realize the Idle-PC values tracked. That's an interesting find. I'll have to test it out.

It's not that it's underpowered, it's that it isn't intended for running stuff like that. I push it to it's max, and I love it for the size/battery life. Great for travelling - except when I expect more out of it than I should.

Specs are: Intel Atom N270 1.6G, 2GB RAM.

I do have a couple of other options. We have a couple of old laptops laying around work that I may be able to use only for GNS3. I also run an esxi server back home that I could possibly build another VM in and use it remotely via VPN. I'll have to check into those options. And I can use this laptop, but I think I will only be able to run a couple routers at once, and even then I've had dynamips crash on me a couple times.


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Wed Oct 05, 2011 10:05 pm 
Offline

Joined: Thu Jun 30, 2011 11:42 am
Posts: 26
Rednectar,

Your solution worked for me big time. I thank you for that. It only need patience and lil bit effort.

HTH
PS


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Wed Jan 04, 2012 3:16 pm 
Offline

Joined: Mon Jan 02, 2012 6:20 am
Posts: 4
I actually disagree with this method.

The best way to acquire Idle-PC is when the router is idle, per the actual dynamips documentation.

You should have no need to set idle-PC after login.

If you set the idle-PC properly, then you will find that you will be able to get low CPU utilization, even without logging into the router.

If possible, please check this idle-PC versus this image, and let me know if this method makes any sense or not (You should have the routers boot up, then have the processes settle to low CPU utilization, even prior to login.)

I have a 12-router topology that will boot up and then come down to 0% CPU within a couple minutes, and this is without opening any console windows.

Please test this IDLE-PC and Image:

idlepc = 0x60bf77a8
image: c3725-adventerprisek9-mz124-15.T7bin

Please reference the idle-PC documentation.

http://dynagen.org/tutorial.htm#_Toc193247999

I don't know when people started doing it wrong, but it seems the "wrong" way got a lot more popular than the correct way of setting this value.

I actually found a post where I guy spoke of the three places idle-PC mattered:

1 - prior to login
2 - after login
3 - after configuration

(Edit: Sadly, I cannot remember where I found that link.)

The best idle-PC would be one that worked effectively across all three scenarios.

Some people have posted that the best IDLE PCs have a 51 or 52 next to them. I have no idea on that, as I never checked. However, I do know that the value given provides great Idle-PC, and I no longer worry about CPU spike after the router has finished booting up, and can boot the entire 12-router topology at once.

With a good idle-PC, the limiting factor becomes RAM.

Hope this helps!


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Wed Jan 04, 2012 7:09 pm 
Offline

Joined: Fri Mar 05, 2010 11:33 am
Posts: 1494
Location: Australia
I don't understand what you don't agree with - the article you refer to http://dynagen.org/tutorial.htm#_Toc193247999 explains precisely the same method as I used, only in less detail and using the dynagen console without GNS3, with the exception that I suggest you set the console timeout to 0 before calculating the idle-pc value.

Nowhere did I say that the router should not be idle. In fact I tell you to conduct the test:
Quote:
When when the router is fully up


Is it the setting of the console timeout that you disagree with? If so, I'd be interested to hear what differences you have found.

To be honest, I've only observed a noticeable difference in CPU utilisation when setting the console timeout when running GNS3 v0.7.x on Windows. I can't say I observed any difference on OS X or Linux when changing the console timeout, and I have a sneaking suspicion that with GNS3 v0.8.x the console timeout is not so critical for Windows either (don't have time to test it now)

Another area I'd also like to test is whether adding modules to your router changes what may have been a good idle-pc into a not-as-good value - especially in relation to the NM-16ESW. Again, I don't have time. Anyone else out there want to take this one up?

_________________
RedNectar
http://rednectar.net
@rednectarchris
GNS3 WorkBench-a VMware image of Ubuntu with GNS3 and VPCS installed and a collection of exercises/labs


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Fri Jan 06, 2012 4:51 pm 
Offline

Joined: Tue Dec 20, 2011 4:05 pm
Posts: 3
I get a really good IDLE PC value - utilization between 0-2%

Then, when I shut down the routers and restart them, the utilization goes straight to 50%

I am using 2 c7200's in my topology. Each one uses precisely 25% utilization, or 50% when both are turned on. It never fluctuates off of that steady 50%.

It does this no matter which IDLE PC value I choose. I have documented them all, and then I go back in and verify in the hypervisor that it indeed is still set with the value that previously delivered perfect results.

Is this expected? Do IDLE PC values have to be calculated upon each boot and/or recycle of the routers? Thanks in advance.

Also, I ventured into the idle pc max /all xxx arena as well. from zero to 1400 and everything in between, nothing drops the 50% utilization down.

I'm on an Intel i5 box, so it doesn't interfere with my other work on the machine. However, when I work from home, this increase in PC utilization causes issues with video and photo loading/buffering. It's quite the nuisance actually!


Top
 Profile  
 
 Post subject: Re: The Illusive Art of Acquiring A Good Idle-PC Value
PostPosted: Tue Jan 31, 2012 6:30 pm 
Offline

Joined: Mon Jan 02, 2012 6:20 am
Posts: 4
rednectar wrote:
I don't understand what you don't agree with - ...
Is it the setting of the console timeout that you disagree with? If so, I'd be interested to hear what differences you have found.


I disagree with logging into the router. I see "logging onto the router" as "doing something", as now, the CPU is probably constantly polling to see if the user enters a command, and is not "idle". This is why when people set the "idle-pc" after they have logged onto the device, that they have to then get to the device console after they turn it on, so that they can "calm down the CPU utilization".

The way the idle-pc works is that it "guesses" when the device shouldn't be doing anything, and saves you some CPU cycles. There is a difference between a device with a logged-in user, and a device "without" a logged in user. There is a greater chance of attaining a more accurate idle-pc value, therefore, if the user is not logged onto the device.

Also, I notice that if you research the ipflow website, when they first came out with the idle-pc feature, it instructed users to do a "wr mem". When you look later at the README's for the 0.2.5 and 0.2.7, you can see that verbiage is removed, and they instead tell you to specifically "NOT" press the enter key, and instead do the key-sequence for calculating idle-pc.


Quote:
To be honest, I've only observed a noticeable difference in CPU utilisation when setting the console timeout when running GNS3 v0.7.x on Windows. I can't say I observed any difference on OS X or Linux when changing the console timeout, and I have a sneaking suspicion that with GNS3 v0.8.x the console timeout is not so critical for Windows either (don't have time to test it now)


It's not important. The important thing is the idle-pc. Usually, the resultant calculation is different for a "logged in user" versus a "not logged in user". Since there are only so many values, you can probably get an occasional collision, where a logged in value is equivalent to a logged out value (had this a couple times during testing), but so far, the best values I get, are when not logged in. (and, they stay good when you log in, configure BGP/OSPF, etc.). I just disagree with telling people to login first ... I see so many prominent, well-known posters advocating logging onto the devices first, such as yourself here, or Martin over at the cisco forums, when I think people would be better off if the idle-pc was calculated outside of logging onto the device. (wouldn't have that mad scramble to log onto the device and start hitting enter, so you could get an "idle-pc")

Even from the documentation, it appears that idle-pc is a "guess" based upon the current state of the router doing nothing. To me, being at an enabled prompt is "doing something". About the best way to workaround that would be what one poster did state on another forum,, was to logout of the device before setting the idle-pc, after setting your exec-timeout 0, ... but as far as making the same config change over and over again ... the software already allows you to specific a default startup-config for your devices.

Quote:
Another area I'd also like to test is whether adding modules to your router changes what may have been a good idle-pc into a not-as-good value - especially in relation to the NM-16ESW. Again, I don't have time. Anyone else out there want to take this one up?


I can report that there is no issue with whatever modules you put into the device.

I use 3725's, and I just model them all the same (I figure the more commonalities between the devices, the more efficiently it can use the memory management features.) I like to use NM-4T, and WIC-2T in all the serial slots. The same image/idle-pc I post above, I use all the time, no issues. I decided to throw a NM-16ESW in one, and it used the same old IDLE-PC, and I had no CPU spiking issues with it at all.

I use the same IDLE-PC across Old Dells running XP 32-bit or Ubuntu 32-bit, and a newer ASUS running Windows 7 64-bit.

CPU is not an issue. I also use idlemax (even though it's probably not even necessary), but since it doesn't introduce any errors for me yet, I'll continue to utilize it. I heard some bad press on "idlesleep" and am not sure if it was ever repaired.

My wife used to have horrendous problems on her computer running gns3, until I gave her my idle-PC to use. It made a world of difference. She didn't even need idlemax anymore.

Hope this helps!




Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 29 posts ]  Go to page Previous  1, 2, 3  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group

phpBB SEO