@coxm Two questions:

What version of fog is this? Is your master node on the same version as the storage nodes?

@coxm Now that you have a working solution, I’ll tell you a bit easier way to go about this.

If you know ahead of time you will keep the OS on a thin image and have the /images on a separate disk then before you install fog create the /images directory and mount your image storage disk over /images and finally install FOG. There will be nothing to change post installation. If you see you need to add the disk after fog is install there are a few more steps involved but that’s basically it.

https://forums.fogproject.org/topic/11048/moving-fog-s-images-files-off-the-root-partition-2017-edition

If you don’t want to disable selinux (or set to permissive) I’ve had luck under CentOS 7 using the following command:

$ sudo setsebool -P tftp_home_dir 1

I thought I might expound a little. To test my Fog TFTP server, I used the TFTP client in Windows. If you would like to do the same, you need to permit TFTP through your Windows firewall.

When I first tried to pull the undionly.kpxe file using the Windows client, I was given the following error:

PS C:\Users\jarcher> tftp -i 172.19.1.30 GET undionly.kpxe Connect request failed

Looking into my /var/log/audit/audit.log file I saw the following:

$ sudo grep tftp /var/log/audit/audit.log ...lots of other stuff here... type=AVC msg=audit(1519225691.090:214): avc: denied { read } for pid=3164 comm="in.tftpd" name="undionly.kpxe" dev="dm-0" ino=34386610 scontext=system_u:system_r:tftpd_t:s0-s0:c0.c1023 tcontext=unconfined_u:object_r:default_t:s0 tclass=file ...also lots more here...

if you use audit2allow you can get some hints on how to fix this error:

$ sudo grep tftp /var/log/audit/audit.log | audit2allow -m tftpd module tftpd 1.0; require { type default_t; type tftpd_t; class file { getattr lock open read }; } #============= tftpd_t ============== #!!!! WARNING: 'default_t' is a base type. #!!!! This avc can be allowed using the boolean 'tftp_home_dir' allow tftpd_t default_t:file { getattr lock open read };

Note, you might have to install the policycoreutils-python package to get audit2allow

Now just run the original command I mentioned above to allow access to the files:

$ sudo setsebool -P tftp_home_dir 1

SELinux is still set to enforcing:

$ sestatus SELinux status: enabled SELinuxfs mount: /sys/fs/selinux SELinux root directory: /etc/selinux Loaded policy name: targeted Current mode: enforcing Mode from config file: enforcing Policy MLS status: enabled Policy deny_unknown status: allowed Max kernel policy version: 28

And now the TFTP transfer works in Windows:

PS C:\Users\jarcher> tftp -i 172.19.1.30 GET undionly.kpxe Transfer successful: 95338 bytes in 1 second(s), 95338 bytes/s

@tom-elliott
Thank you for your reply. I thought about this too but I’m a bit uncomfortable with modifying the FOG code. Plus, it would break if I upgrade.
I got another idea yesterday, after writing this post. I was thinking about a script, that would run from cron, about every 10 minutes. It would check if udp-sender is running. If it isn’t, it would just exit. If it is, it would start monitoring the process and, when it finishes, it would launch my custom code.
This would work if udp-sender runs only during multicast. I suppose this is the case. I don’t see any reason for it to run otherwise.
How does this sound?

@julianh said in Running out of Space during Imaging:

can I redirect the /images/dev to another disk?

Yes, you can do that. But there are some other questions to ask to figure out why you are having these problems and if there are any obvious quick wins we can help you with:

What is the output of df -h Output of lsblk Is this in a VM or a physical machine? Do you have more than one spare disk you can use for the FOG Server?

@sebastian-roth

I just happen to be harvesting 1TB drives from workstations for server use. The 250GB drives I picked up for $10 each are adequate for the floor computers.

Thank you for the firm point at cause.

I’m testing your idea but the msi is very badly done ! 😞

@szeraax Thanks for the update. It’s definitely an issue that we will be seeing more and more in the next month.

Not because it is solved, but because I will just have to wait for a magical time in the future when FOG has support for booting NVMe hard drives decently.

Unfortunately those things rarely solve themselves magically. People have to work on it. In this case having the hardware at hand seems to be very crucial as your findings reveal (same zbook model showing different behavior). By the way, have you checked firmware versions?
So what I am trying to say is that it might need your work on testing these things on your devices to hopefully figure things out. But on the other hand I do understand that you/we don’t have the time and expertise to dig through this.

Maybe try going back to using rEFInd plus adding the clover NVMe driver? As I understand it you should be able to add that driver by simply creating a directory drivers_x64 (EFI/boot/drivers_x64) and putting that clover driver EFI file there. rEFInd should scan that directory and load the driver.

Not sure if we’d mark this solved.

Of course - checked the logs and I had the user name or password in wrong in fog. Had to be easy. Thanks!

So I looked at the file with my wireshark at home (forbidden in office)
There´s no data.

I will set up a new fog from scratch and try to WOL the devices. I will report later.

thanks Wayne. I will do this this weekend and if I need help post back here.

@wayne-workman said in Image not syncing:

As far as the “error contacting server” message, can you visit https://sourceforge.net/ at your work? If not, that’s probably why.

I can access it on my network. I have perform below command in order to try and fix the syncing issue. Than I re-ran the fog installtion file to re-install. When I was checking the master fog storage under storage. I noticed password for the user fog was simply password and wasn’t matching the password in the .fogsettings file.

sudo mv /opt/fog/.fogsettings /opt/fog/fogsettings-firstInstall

Perfect! That worked a treat.

In the interest of housekeeping, can I now delete that group? If so do I need to remove hosts first?

Thanks,
Matt

@sebastian-roth @Tom-Elliott
Thank you both so much, through a combination of things I have it working. you guys ROCK!

Thanks again!

@sebastian-roth I was referring to my intended use case. I want to capture an image from a computer and automatically copy the individual files from the image onto a file server. In order to do that, I need a script that runs on the server, not the host. I could find a way to trigger it still (some primitive form of RPC, maybe a quick HTTP request with wget), but it wouldn’t be quite as convenient.

@tom-elliott I am actually thinking it might have been a BIOs setting. It was on RAID mode so I reinstalled Windows and just captured it and it appears to have booted back into my Windows this time.

I’ll finish up this install and try again and let you know.
Thanks!

@techadmin Yes you can delete those. Also you should read this: https://wiki.fogproject.org/wiki/index.php?title=Troubleshoot_MySQL#Repair_broken_database

@UWPVIOLATOR This has been fixed in v0.11.14, which will be released with the next server RC. I’d still recommend using process-explorer to find out why the file is locked though.

To get v0.11.14 you can either wait for the next server release, or download it from here: https://github.com/FOGProject/fog-client/releases/tag/0.11.14

@bigchief Sounds like a mismatch of bios/ueif to me. A UEFI system can sometimes have the feature to optionally legacy boot (bios). What mode is the firmware in? (bios/legacy or UEFI)? So figure that out - and then dig through the firmware settings to make sure no optional Legacy modes are enabled.

Also, could be something as simple as turning off the HDD’s RAID mode in firmware and enabling AHCI mode.

@warp said in WoL and VLAN:

On the other hand a tool of WoL Windows works well !

Maybe that’s from a different subnet/vlan? Can you try from a Windows machine in the same subnet as your FOG server is, so 172.16.8.x! If that works than we can start looking into FOG being the issue.