Network booted
-
We're imaging 1000 computers this month. Thanks to fog, it's pretty easy. I WoL the whole lab into Linux and they image automagically.
Until students step in.
Last year the summer cleaners hooked a patch cable into two drops to keep it out of the way. The switch wasn't very smart, and the whole network went down. I spent an hour finding the loop and we updated the permanent switches after that.
This summer, the tech help wired the wireless clients to temp switches, but they wouldn't network book. The spinning bar (/-\|) didn't even spin.
They blamed the server, and I started checking for loops in the cables. 1st switch, no cable leaving and returning, 2nd switch, same, ect.. They had made a mesh topology. I need smarter switches or smarter help.
One particular Heaping Pile desktop refuses to network boot. The first workaround was gPXE by CD, but then the kids rip out the drive belts.I don't feel like replacing the drives, and I get tired of using two tweezers and a rubber band to fix it for a week until they pull it off.
I tried USB gPXE, which was booting to linux to boot from the ethernet to boot linux to install Windows, but that wasn't complicated enough, so now I use RMprepusb to load linux from the USB to choose an ISO to boot linux to boot from the ethernet to boot linux to install Windows.
-
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
-
@trainbrain27 said:
We're imaging 1000 computers this month. Thanks to fog, it's pretty easy.
I've never seen Fog do proper multicasting so it's basically a big queue. I guess if you don't have a decent budget it's ok but the delivery time compared with commercial solution is terrible.
@trainbrain27 said:
Last year the summer cleaners hooked a patch cable into two drops to keep it out of the way. The switch wasn't very smart, and the whole network went down
That's not a WTF. Closing a loop and doing a power-cycle is a common way to reset a switch.
@trainbrain27 said:
This summer, the tech help wired the wireless clients to temp switches, but they wouldn't network book. The spinning bar (/-|) didn't even spin.
They blamed the server, and I started checking for loops in the cables. 1st switch, no cable leaving and returning, 2nd switch, same, ect.. They had made a mesh topology. I need smarter switches or smarter help.
One particular Heaping Pile desktop refuses to network boot. The first workaround was gPXE by CD, but then the kids rip out the drive belts.I don't feel like replacing the drives, and I get tired of using two tweezers and a rubber band to fix it for a week until they pull it off.
I tried USB gPXE, which was booting to linux to boot from the ethernet to boot linux to install Windows, but that wasn't complicated enough, so now I use RMprepusb to load linux from the USB to choose an ISO to boot linux to boot from the ethernet to boot linux to install Windows.
If you plug/unplug cables like a 1949 AT&T switchboard operator there is no wonder you have bad network performance. Most switches will build an internal ARP dispatch table and will not forward packets between two different ports if according to their manifest the source and destination are on the same segment. If you move nodes from one port to another one and you can't afford a more sophisticated edge you will experience bizarre behaviors.
-
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
I imagine he means that the nice little children are too impatient to wait for the CD to eject and just rip the drive open. ...But that'd require belt-geared drive opening mechanisms, which I have yet to see.
It's more common anyways for the drive to spin the disc platter with a belt. In laptop-type cd-drives, you can usually see the belt. I presume that in the Lowest-Bidder, Inc. model the belt is not protected by anything.
...Or these might just be smart kids who just happen to enjoy tormenting others. What's the word... precocious? And they've obviously learned hod to rip a drive-belt out through the narrow gap between the tray and body with a pencil. ...Is it a cynecology school, by any chance?
-
@OldCrow said:
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
I imagine he means that the nice little children are too impatient to wait for the CD to eject and just rip the drive open. ...But that'd require belt-geared drive opening mechanisms, which I have yet to see.
It's more common anyways for the drive to spin the disc platter with a belt. In laptop-type cd-drives, you can usually see the belt. I presume that in the Lowest-Bidder, Inc. model the belt is not protected by anything.
...Or these might just be smart kids who just happen to enjoy tormenting others. What's the word... precocious? And they've obviously learned hod to rip a drive-belt out through the narrow gap between the tray and body with a pencil. ...Is it a cynecology school, by any chance?
Smash their fingers with a crowbar and then make them fix whatever drives they broke, the little asses.
-
@morbiuswilters said:
@OldCrow said:
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
I imagine he means that the nice little children are too impatient to wait for the CD to eject and just rip the drive open. ...But that'd require belt-geared drive opening mechanisms, which I have yet to see.
It's more common anyways for the drive to spin the disc platter with a belt. In laptop-type cd-drives, you can usually see the belt. I presume that in the Lowest-Bidder, Inc. model the belt is not protected by anything.
...Or these might just be smart kids who just happen to enjoy tormenting others. What's the word... precocious? And they've obviously learned hod to rip a drive-belt out through the narrow gap between the tray and body with a pencil. ...Is it a cynecology school, by any chance?
Smash their fingers with a crowbar and then make them fix whatever drives they broke, the little asses.
Better to smash their knees first when you force them to fix it, then smash the fingers afterwards. That way they still have full manual dexterity while fixing the drives.
-
@DescentJS said:
Better to smash their knees first when you force them to fix it, then smash the fingers afterwards. That way they still have full manual dexterity while fixing the drives.
That's the point: make them put it back together with smashed fingers. If they don't do it fast enough, then we smash knees!
-
@morbiuswilters said:
@DescentJS said:
Better to smash their knees first when you force them to fix it, then smash the fingers afterwards. That way they still have full manual dexterity while fixing the drives.
That's the point: make them put it back together with smashed fingers. If they don't do it fast enough, then we smash knees!
Good point, but I didn't say the finger smashing was optional. They would always get both smashed fingers and knees, you just don't have to wait as long for them to fix the computer.
-
It's a High School. I assume that refers to the condition of the students.
This is very similar to our drives.
-
@Ill Stew said:
It's a High School. I assume that refers to the condition of the students.
You might like to think about whether, as is typical with schools, you've set the computer labs up as what amounts to a challenge to the students to find creative ways to break things. I've never had to manage kids, but with some less mature and intelligent adults I've had good results by giving them a little more responsibility. A few idiots will always exploit anything you give them, and have to be locked down, but most people respond well to being treated like adults: you want to reverse the normal school paradigm where kids are used to being watched and told to stop doing stuff if it's not allowed.
I've been known to tell people in business that they'll have better results if they stop treating their employees like kids in school, but you're actually dealing with kids in school. I wonder if it still works.
-
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
I assume that they remove the drive belt that opens the optical drive trays. It is easily accessible when the drive is opened. Whenever I have to use an optical drive this belt has stretched and I need to assist the tray to open with a paper clip.I suppose they have some use for a 1¼" square profile drive belt.
-
Ronald, I don't particular care about you constantly changing the image in your signature, but I admit this one is kinda funny.
-
@Ronald said:
@trainbrain27 said:
Last year the summer cleaners hooked a patch cable into two drops to keep it out of the way. The switch wasn't very smart, and the whole network went down
That's not a WTF. Closing a loop and doing a power-cycle is a common way to reset a switch.
Designing a switch that resets (I assume you mean to factory defaults) when faced with one of its most likely (and trivial) misuse scenarios seems a little short-sighted...
-
@Hatshepsut said:
@Ronald said:
@trainbrain27 said:
Last year the summer cleaners hooked a patch cable into two drops to keep it out of the way. The switch wasn't very smart, and the whole network went down
That's not a WTF. Closing a loop and doing a power-cycle is a common way to reset a switch.
Designing a switch that resets (I assume you mean to factory defaults) when faced with one of its most likely (and trivial) misuse scenarios seems a little short-sighted...
Commercial switches are not like LAN party netgears where people keep putting cables in an out. Usually in a typical data center once the patching is done it stays like that; servers are connected to patch panels and lan drops go all over the building; most activity occurs at the outlet level. Pulling cables all the time is usually a sign that there has been no planning or that there is no skilled staff to tend to the network.
-
@Ronald said:
@Hatshepsut said:
@Ronald said:
@trainbrain27 said:
Last year the summer cleaners hooked a patch cable into two drops to keep it out of the way. The switch wasn't very smart, and the whole network went down
That's not a WTF. Closing a loop and doing a power-cycle is a common way to reset a switch.
Designing a switch that resets (I assume you mean to factory defaults) when faced with one of its most likely (and trivial) misuse scenarios seems a little short-sighted...
Commercial switches are not like LAN party netgears where people keep putting cables in an out. Usually in a typical data center once the patching is done it stays like that; servers are connected to patch panels and lan drops go all over the building; most activity occurs at the outlet level. Pulling cables all the time is usually a sign that there has been no planning or that there is no skilled staff to tend to the network.
Sure. But requiring to patch a switch to itself to restore defaults (and to disconnect it afterwards) seems clunkier and more error prone than, say, pushing a recessed button dedicated to that purpose.
-
@trainbrain27 said:
We're imaging 1000 computers this month. Thanks to fog, it's pretty easy. I WoL the whole lab into Linux and they image automagically.
Until students step in.
Last year the summer cleaners hooked a patch cable into two drops to
keep it out of the way. The switch wasn't very smart, and the whole
network went down. I spent an hour finding the loop and we updated the
permanent switches after that.This summer, the tech help wired the wireless clients to temp switches, but they wouldn't network book. The spinning bar (/-|) didn't even spin.
They blamed the server, and I started checking for loops in the cables. 1st switch, no cable leaving and returning, 2nd switch, same, ect.. They had made a mesh topology. I need smarter switches or smarter help.
Then again, a few decades back, spanning tree was invented to cope with loops and meshes - in fact it changes meshes from non-working to being a resilient system
-
@jes said:
Then again, a few decades back, spanning tree was invented
To mangle a quote:
"A network engineer, when faced with a problem, thinks 'I know, I'll enable spanning tree!'. Now he has two problems."
But yeah he should have STP on those switches, unless they're unmanaged or cheap top-of-rack type switches, in which case he should just burn the lot to the ground and go and live in a cave somewhere.
-
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
Yeah, WTF...?!?
-
@robbak said:
@morbiuswilters said:
@trainbrain27 said:
The first workaround was gPXE by CD, but then the kids rip out the drive belts.
Huh?
I assume that they remove the drive belt that opens the optical drive trays. It is easily accessible when the drive is opened. Whenever I have to use an optical drive this belt has stretched and I need to assist the tray to open with a paper clip.I suppose they have some use for a 1¼" square profile drive belt.
TRWTF is that there is a belt in the drive opening mechanism. The last drives I pulled part for fun and small DC motors, from somewhere around year 2000, had hard plastic gears only. They managed to outlive their host PCs. The belts... well... they ended up on TDWTF.
Possibly, the TRRWTF is putting the belt in a place where it can easily be sabotaged, when the repair instructons specify that the drive needs to be partially disassembled anyway.
-
@OldCrow said:
The last drives I pulled part for fun and small DC motors, from somewhere around year 2000, had hard plastic gears only.
The drive in my 2 years old HP has a belt that can be accessed when the tray is open:
