Using dumb switch to troubleshoot imaging, network related

[Wayne Workman]

@Mastriani Stop down-talking yourself.

apt-get install iperf ethtool -y

apt-get is Ubuntu’s preferred package manager. There is also apt and dpkg

install is one of many commands to do. There is also remove and others.

iperf and ethtool are package names.

-y means do whatever needs done to just install it and don’t ask for permission, this is my permission here.

Fedora 24’s package manager is dnf but the rest is the same. In CentOS 7, it’s yum. In Arch Linux, it’s pacman

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani Stop down-talking yourself.

apt-get install iperf ethtool -y

apt-get is Ubuntu’s preferred package manager. There is also apt and dpkg

install is one of many commands to do. There is also remove and others.

iperf and ethtool are package names.

-y means do whatever needs done to just install it and don’t ask for permission, this is my permission here.

Fedora 24’s package manager is dnf but the rest is the same. In CentOS 7, it’s yum. In Arch Linux, it’s pacman

Objectively, better down than up, and this is 3 years of frustration speaking, having attempted to learn and resolve this solo. I will post back when the necessary conditions have been met. Thank you.

[Tom Elliott]

@Mastriani The UDP issues might have come up due to 0.29. The NFS mount was not using the TCP proto (of which it does now).

[Wayne Workman]

I’ll also be looking at some of the same stuff that George already touched on. And I’ll of course be using the same “near to far” thing he was talking about, this is just how to properly troubleshoot.

[Mastriani]

@Tom-Elliott said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani The UDP issues might have come up due to 0.29. The NFS mount was not using the TCP proto (of which it does now).

The current servers are clean install of FOG 1.0.2 / Ubuntu 14.04 LTS, because other members of the team had targeted the old server as being the issue, as opposed to the change-over of network hardware. It was a post hoc ergo propter hock situation, having also migrated from FOG .32 / Ubuntu 10.04 to FOG 1.0.1 / Ubuntu 12.04. This was done during the network hardware change-over.

I doubted the correlation as causation, the only issue with 12.04, (appears to be fairly common), was installing the Unity desktop environment. Which was later removed and replaced with LXDE, preferable to my mind.

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

I’ll also be looking at some of the same stuff that George already touched on. And I’ll of course be using the same “near to far” thing he was talking about, this is just how to properly troubleshoot.

I will most certainly appreciate the learning process and your time.

[Mastriani]

@george1421 said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani Well the 755 to 390s are not the most powerful systems but they should not take 30 minutes to image either.

When you run you test, please try to get an accurate time to deploy and the actual image size. That way we can calculate the true throughput. The number from part clone is a bit deceiving.

The compression value is a scale from 0 (no compression) to 9 (maximum compression). That also means:
0 = maximum file size on the FOG server and over the network but images faster since the client doesn’t need to decompress the image.
9 = smaller file size on the FOG server and network but images slower since the client must squeeze all of the air possible out of the image to capture and deploy it.

5 is a middle of the road between larger file and higher cpu requirements on the client.

FOG server and client machine on same switch. Image capture started at 10:28 a.m. EDT
Dell OptiPlex 780
Dual Core
4Gb PC3 DDR RAM
GbE Broadcom NIC, 1 Gb/s as per device manager
34.3 Gb image
Started: 2.8 Gb/s, 10 mins. into capture, speed degraded to 1 Gb/s, 11%, if the partclone measurement is believable.
20 mins, 620Mb/s, 18%
30 mins, 450Mb/s, 39%
40 mins, 395Mb/s, 49%
50 mins, 325Mb/s, 57%
60 mins, 315Mb/s, 61%
70 mins, 295Mb/s, 65%
80 mins, 272Mb/s, 69%
90 mins, 261Mb/s, 72%
100 mins, 250Mb/s, 77%
110 mins, 237Mb/s, 81%
120 mins, 223Mb/s, 88%
130 mins, 215Mb/s, 93%
140 mins, 222Mb/s, 97%
148 mins, 209Mb/s, 100%
Total time: 2 hr, 31 mins, 34.3 Gbs transferred, 12:59 p.m. EDT

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

I’ll also be looking at some of the same stuff that George already touched on. And I’ll of course be using the same “near to far” thing he was talking about, this is just how to properly troubleshoot.

iperf and ethtool installed on FOG server, temporary interruption for educator request, will setup the laptop on the same switch with live disc, shortly.

[Wayne Workman]

@Mastriani When you’re ready, send me an IM. Forums posts are too slow for this sort of help.

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani When you’re ready, send me an IM. The forums are too slow for this sort of help.

FOG server is Ubuntu 14.04 LTS, match version on disc, or not relevant?

[Wayne Workman]

@Mastriani use 16.04 for the live disk, there’s no reason not to.

[Mastriani]

@Tom-Elliott said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani The UDP issues might have come up due to 0.29. The NFS mount was not using the TCP proto (of which it does now).

Mr. Elliott,

Does FOG use jumbo frames? Fixed rate, negotiated, or not relevant?

[Tom Elliott]

@Mastriani I’m fairly sure (unless you set your nic’s up for jumbo frames) it’s irrelevant.

[Quazz]

I do recall something about negotiated being optimal, but nothing beyond that.

[Wayne Workman]

This was due to a known issue with the onboard NIC chipset the server had.

@Mastriani could you post the chipset model please? I forgot what it was.

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

This was due to a known issue with the onboard NIC chipset the server had.

@Mastriani could you post the chipset model please? I forgot what it was.

Both of the Broadcom NIC’s used in Dell servers are an issue, what is now known thanks to your workup on my server Mr. Workman.

Broadcom NetXtreme 5721 Single Port Gigabit Ethernet NIC, is what is in mine.

Hands down, Fog, it’s development team and support professionals, are inarguably in a class of their own. My sincerest thanks and professional appreciation to all, Mr. Elliott, and a note of exceptional knowledge and top class support and effort to Mr. Workman.

There is in my estimation, none better. Were it that others in the IT community would take lessons from your organization and team. Well done.

[Wayne Workman]

@Mastriani So, you put in another NIC, configured it, disabled the others, and the issue is solved?

[Mastriani]

@Wayne-Workman said in Using dumb switch to troubleshoot imaging, network related:

@Mastriani So, you put in another NIC, configured it, disabled the others, and the issue is solved?

Currently capturing a 40Gb image, total time less than 17 mins.

Yesterday, updated 2 labs of 30 machines each, in 2:45:00, from start to finish.

The Trunk version is smooth, intuitively easy, and scorching fast. All is well for a small rural district in SE Tennessee!

Well, except that it is SE Tennessee.