FOG 33b - installation - HP Elite 8300 SFF
-
[code]
sudo service tftpd-hpa stop
sudo service tftpd-hpa start
[/code]Is what you should be using.
You are lucky "sudo stop tftpd-hpa " even works at all, this is not how you syntax a service restart.
You have confirmed that there is a bug in the tftpd-hpa up start service, this is a known issue where the service is started before an ip address is even assigned to the machine.
[B]If I were you I would manually type the commands at the restart of the server.[/B]
This is not a problem of fog, your fog server is working as expected. If you want help in resolving the issue, please try an Ubuntu forum, the people there are very nice and will help you to better understand the problem you are facing and how to fix it!
-
I repeated the above procedure only one time and multicast failed to my horror.
This failure of multicast is totally unpredictable.
Jaymes, the syntax sudo stop tftpd-hpa does in fact work. It is confirmed on the following line that tftpd is not running.
But I will use the syntax you recommend from now on.You wrote: [B]If I were you I would manually type the commands at the restart of the server.[/B]
A have done that [U]everytime[/U] as I listed above. Multicast works sometimes and not others.
This is crazy. I had the TFTP problem with the old fogserver and old HP clients. But that was easily fixed with the start stop syntax.
Never had a problem with multicast, if that is in fact the real problem here. -
Why multicast then? Try unicast to multiple hosts, I personally hate using multicast because if a host drops out in the middle of imaging, your whole group is hosed. I set up a group of members and push a unicast to each of those machines at once. only 10 will image at a time but they will wait in que until a spot opens up and then image.
Try unicast imaging, it uses the tftp-hda service to push to single units too, try it before you multicast.
So reboot, try single unit image, is it successful?
If not, and it is because of the service, restart it and push image again, was it successful?after successful unicast, try multicast, did it fail? Try unicast again, did it succeed?
what I’m trying to do here is figure out if it is the tftpd-hpa service, or if it is just the multicasting ability.
Sorry to give you so much to do, but I want to get some direction going here. Is the multicast only failing for these units? Are you still working on a test switch setup like I suggested? Try these troubleshooting steps I recommended OFF of the network, so we can make sure the FOG server is working.
We want to get multicast for SURE working before you put it in the production environment because some switch gear causes issues.
-
[quote=“Alistair, post: 25671, member: 1830”]TFTP is still not running despite the mod to /etc/rc.local
Here is a copy of the file. Is there anything wrong?
#!/bin/sh -e
rc.local
This script is executed at the end of each multiuser runlevel.
Make sure that the script will “exit 0” on success or any other
value on error.
In order to enable or disable this script just change the execution
bits.
By default this script does nothing.
/bin/sleep30 && /etc/init.d/tftpd-hpa restart
exit 0xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
In order to enable or disable this script just change the execution
bits. (no idea what this means. Do I need to enable this script?)[/quote]
I believe your script has a typo:
I should read as:
[code]
#!/bin/sh -erc.local
This script is executed at the end of each multiuser runlevel.
Make sure that the script will “exit 0” on success or any other
value on error.
In order to enable or disable this script just change the execution
bits.
By default this script does nothing.
/bin/sleep 30 && /etc/init.d/tftpd-hpa restart
exit 0[/code]Major difference is that there was no space between sleep and 30 originally so it read: sleep30, I made it read sleep 30.
-
[U]I am back on the simple one 4port router network with:[/U]
Server 2012 (for DHCP)
fogserver (I am running management console here)
2 clients
I pinged router and server - both respond.Rebooted fogserver
stop start tftpd
Multicast to 2 PCs
Multicast not work
Cancel tasks
Unicast to 1 PC
Worked.
Multicast to 2 PCs
Multicast not workThanks Tom for the typo correction - To avoid a typo I copied the syntax directly from:
[url]http://fogproject.org/forum/threads/ubuntu-11-04-distro-that-just-works-heres-how.4378/#post-12571[/url]
I just cannot win here. -
Just because you have multiple systems, does not mean you have to Multicast to get a job to work. You can schedule both systems as Unicast tasks, even if they have the same image and information.
Beyond that, what kind of router are you using? My home network goes completely dead if I try to multicast to two systems at the same time as the Router I have at home is a typical Consumer router.
I imagine it’s the same issue being experienced here.
-
[quote=“Alistair, post: 25686, member: 1830”]
Thanks Tom for the typo correction - To avoid a typo I copied the syntax directly from:
[url]http://fogproject.org/forum/threads/ubuntu-11-04-distro-that-just-works-heres-how.4378/#post-12571[/url]
I just cannot win here.[/quote]That one was my fault there was no space between sleep and 30, I have fixed the post so the problem shouldn’t be present in the future.
Alright so now we have determined the problem to be with the multicast directly. So here’s something for you to ponder, it is NOT the tftpd-hpa service, if it were, unicast would fail, and any attempts to use a command to communicate with the tftp server would fail.
WITHOUT REBOOTING OR DOING ANYTHING ELSE, while your unicast works and multicast doesn’t, use the tftp get command on a windows machine on the 4 port switch.
use this command
[code]
tftp IPADDRESSOFFOG get undionly.kpxe
[/code]If the command succeeds. It is not a problem of the tftpd-hpa service, it may be the hardware you are using. If you are successful, and you can image the machines via unicast, please try a unicast group to see if you are able to achieve imaging multiple machines. If successful, look into a business class switch or hub and try multicasting again.
Multicast is not supported by all hardware.
I don’t want to sound pompous, but this may not be an issue of FOG, this may be an issue of the gear you are using.
-
Still on the simple 1 router network.
A unicast to a group of 2 PCs has worked.
tftp 10.0.1.40 get undionly.kpxe
“tftp not recognised as an internal command” -
tftp client is not installed by default in windows, add it in the “turn windows features on or off” section of “programs and features”
-
[quote=“Alistair, post: 25689, member: 1830”]Still on the simple 1 router network.
A unicast to a group of 2 PCs has worked.
tftp 10.0.1.40 get undionly.kpxe
“tftp not recognised as an internal command”[/quote]I assume windows 7?
The tftp command is not available by default in Windows 7 but can be enabled by turning on the TFTP Client Windows feature from Programs and Features in Control Panel.
That is my fault, I still have Windows xp floating around that I use for my virtualized images, I usually run commands from there, I had forgotten that the latest revisions of windows don’t have tftp enabled by default.
-
This simple network router is a domestic wireless router that I have lying around.
My main system router is a new expensive PaloAlto Firewall which is the gateway for 6 vlans.
But as I said earlier, all of these new and old HP PCs, server, fogserver etc are on only 1 vlan.
There is no cross vlan interaction. There are no multiple systems that concern us here. Just one.
The PaloAlto should have no problem at all in handling multicast.
The simple domestic router I’m using right now did in fact work with multicast yesterday after I rebooted fogserver.
My central switch is a Cisco SG-300-52 port gigabit.When multicast is working, [U]it appears to keep on working for multiple jobs until I reboot fogserver.[/U]
When multicast is not working, it never works again without rebooting. -
tftp client is now installed on win 8.1 client in Audit mode (one of the PCs that has been receiving the images)
tftp 10.0.1.40 get undionly.kpxe
flashing cursor…
“Connect request failed”I’ll try turning off firewall
-
[quote=“Alistair, post: 25693, member: 1830”]tftp client is now installed on win 8.1 client in Audit mode (one of the PCs that has been receiving the images)
tftp 10.0.1.40 get undionly.kpxe
flashing cursor…
“Connect request failed”I’ll try turning off firewall[/quote]
yes please disable the firewall it causes issues.
I read your posts, I understand you are using that Firewall, that is fine but for now we are trying to eliminate it as a possibility of the errors, so eliminate extra switch gear use only the router. I understand your set up, that is fine but first lets get for WORKING then lets introduce it to the network. Adding Servers is like adding fish to your aquarium, you need to make sure that the “fish” get use to the temperature of you water, and then you can relase them, same thing here, lets let FOG get use to being in the workplace and working then lets release it to the production environment. If the router worked yesterday for multicast, then it will work today, we just have to figure out what is failing and why.
Why are you rebooting the server after you get it working are you a glutton for punishment?
on a side note I am going to test multicasting.
-
With firewall off:
“Transfer successfull”
-
[I]now we are trying to eliminate it as a possibility of the errors, so eliminate extra switch gear use only the router.[/I]
I have been ONLY using one router, no other switches or firewall.[I]if the router worked yesterday for multicast, then it will work today, we just have to figure out what is failing and why.[/I]
It is totally unpredictable. This is the problem.[I]Why are you rebooting the server after you get it working are you a glutton for punishment?[/I]
This is an important principle of technical fault finding. Always try to reproduce exactly the same circumstances under which something happened or didn’t happen. A fresh reboot gives you that. -
This post is deleted! -
[quote=“Alistair, post: 25695, member: 1830”]With firewall off:
“Transfer successfull”[/quote]
This is not an issue of tftpd-hpa.
If you are interested in getting rid of the tftpd-hpa upstart as being a POSSIBLE issue you can install an older version of Ubuntu, I recommend 10.04
I tested multicast on the latest FOG revision, Multicast works.
you can try turning off multicasting service and turning it back on you can do so with the following command
[code]
sudo /etc/init.d/FOGMulticastManager stop
sudo /etc/init.d/FOGMulticastManager start
[/code]Then try multicasting.
-
Jaymes - Thanks for those final tips on stopping and starting multicast.
I’ll try that. I’m not going back to 10.04.[FONT=Times New Roman]I can’t spend any more time on this. I’m kicking a dead horse. I’m running with unicast. I know it’s slower and only 10 at a time. I have no more time to use on this multicast problem. I’m usually good at fault finding, but here I have failed to find any pattern of success and failure. One time in another school I had 50% of students who could print to a particular printer and 50% who could not. I eventually discovered that all those who had domain user names longer than 15 characters, could not print out on that particular printer. No problem in printing to other printers. Go figure. Over 20 years I’ve seen [U]many[/U] examples of 1 program creating a problem in another on various platforms. Drivers for smartboards screw up LibreOffice. [/FONT]
[FONT=Times New Roman][/FONT]
[FONT=Times New Roman]I suspect there is a problem in fog, maybe connected with firmware on my new HP Elite 8300 SFF and USDT clients.[/FONT]
[FONT=Times New Roman]I’m running fog 33b on one of the new SFF. But I also think it is strange and disappointing that I cannot unicast or multicast to my old HP DC7100 clients that I imaged many times from my old fogserver version 0.32.[/FONT]Thank you Tom, Jaymes and JunkHacker for all your input and time.
I will only give further comments in this thread if I discover something significant that you should know about.[FONT=Times New Roman][/FONT]
-
[quote=“Alistair, post: 25701, member: 1830”]Jaymes - Thanks for those final tips on stopping and starting multicast.
I’ll try that. I’m not going back to 10.04.[FONT=Times New Roman]I can’t spend any more time on this. I’m kicking a dead horse. I’m running with unicast. I know it’s slower and only 10 at a time. I have no more time to use on this multicast problem. I’m usually good at fault finding, but here I have failed to find any pattern of success and failure. One time in another school I had 50% of students who could print to a particular printer and 50% who could not. I eventually discovered that all those who had domain user names longer than 15 characters, could not print out on that particular printer. No problem in printing to other printers. Go figure. Over 20 years I’ve seen [U]many[/U] examples of 1 program creating a problem in another on various platforms. Drivers for smartboards screw up LibreOffice. [/FONT]
[FONT=Times New Roman]I suspect there is a problem in fog, maybe connected with firmware on my new HP Elite 8300 SFF and USDT clients.[/FONT]
[FONT=Times New Roman]I’m running fog 33b on one of the new SFF. But I also think it is strange and disappointing that I cannot unicast or multicast to my old HP DC7100 clients that I imaged many times from my old fogserver version 0.32.[/FONT]Thank you Tom, Jaymes and JunkHacker for all your input and time.
I will only give further comments in this thread if I discover something significant that you should know about.[/quote]
I don’t know what your fear of 10.04 is, you could do so as a test, then install latest fog and try multicasting, if multicast works then there may be a problem with something installed on FOG server A (The one with the broken multicast).
You can increase the number of clients that can unicast at once, you can change this number under the fog settings of the web gui, click on General Settings and then edit the que size.
but WHAT gives you the impression that it is slower? Unicast images the machines one at a time and decompresses the data AT the host. with multicast you decompress at server and send ALL information across the network, and then you wait for all multicast hosts to get to the same spot then they move forward. Multicast is slower than unicast… AND has the potential to leave you without an OS should you choose to leave while Multicast is performing.
You can continue to blame FOG if you want, but the issue is specific to your server at the moment, there are many of us that use FOG for multicast, I do not I find more faults in the system than good, and they are not reporting this issue.
I HAVE in the past 30 minutes, multicasted a number of machines without any problems, the biggest problem I faced was sitting at PLEASE WAIT and I was able to solve this by restarting the Multicast Manager after I changed the Multicast Max sessions on the fog web GUI to another number and back again and saving the page.
I wish you the best of luck. You can still unicast to groups and deploy snap-ins you are not at a total loss because multicast is not working properly.
-
Jaymes has found the problem!
sudo /etc/init.d/FOGMulticastManager stop
Stopping fog imaging solution: FOGMulticastManager
start-stop-daemon: warning: Failed to kill 1157: no such processsudo /etc/init.d/FOGMulticastManager start
Starting FOG solution: FOGMulticastManagerMulticast worked !!!
“no such process” !!!
It wasn’t running.
