Posts made by Junkhacker

fog works by mounting it’s storage on the systems it’s imaging as an nfs share. you would be exporting a mounted nfs share over nfs… systems don’t like that. and you have to modify the default (sane) settings to allow it to happen.

@brakcounty that would not alter that screen. that would display “Hello <contents of first field>” on the next screen after hitting enter

@jra correct

@jra you don’t

no longer needed. it wasn’t very good encryption when we did use it either.

@george1421 said in OS Support - the numbers are in:

Finally on my campus we are slowly moving away form Centos to Debian as the platform of choice for new installs of linux.

Just wanted to chime in that my campus is almost exclusively Debian as well. except fog, because i installed it on Ubuntu when i was just getting started, and because Ubuntu is for beginners*.

Personally i have preferred Debian over CentOS because of the incredibly stable Enterprise Linux that is RHEL/CentOS has always been so slow to get performance/feature enhancements.
I prefer Debian over Ubuntu because it doesn’t break things as often changing things for the sake of change, lol.

*(joking, obviously. mostly.)

log4j is a java library that is sometimes packaged with a program without it being installed on it’s own, so it doesn’t have to show up in the apt list to be present. that said, george is right that fog doesn’t use java, so you don’t have to worry about it there.

@richard-wise if you go the route of using an external csv file, might i suggest putting it in the /images/postdownloadscripts directory of the fog server? it’s a location that will be mounted anyway

@g-sivakolunthu while possible, it’s not what fog is made for and you would have a bunch of manual steps. you might want to look at clonezilla for imaging with online images possibly in combination with gparted for partition resizing.

@florent in my alternate setup, only the machine you use to upload images to fog would need NFS enabled. the hosts you deploy to would not.

@george1421 i have had in the past fog working without NFS completely (for downloads. uploading images to fog still required NFS, but that’s easier to lock down since you can limit access by IP or whatever for your image source machine)

if there’s any interest, i have ideas on how to re-implement fogtorrent (fog imaging using bit-torrent) though my time to participate in it’s developent (and skill level) are limited/

@brakcounty i don’t know what’s going on, but make sure you don’t have any typos and that you’re running the script with python3

@brakcounty from a bit of googling, it looks like maybe something changed in the module. where i have “username” you should replace with “user”

@florent securing fog has always been one of the biggest challenges. there isn’t a good way to secure postdownloadscripts while still allowing them to be accessible by FOSS. what could be done if you have anything requiring some level of security would be for the script to not actually be what you need done, but a reference to an external source for what you need. a https address for example. the files you’re serving up from that source could be secured and access restricted based on what the FOSS knows or the FOG server knows (i.e. host with this ID in the system and an active imaging task is allowed download access)

@brakcounty this is definitely a VirtualBox issue, not a fog issue, but as i recall VirtualBox uses a version of iPXE as it’s built in network boot rom. maybe you can directly swap the binary with an updated one?

@brakcounty find the password in the config file fog uses to connect to the database.

make sure the directory you told it to use is correct. compare it to the other working images. and remember, it is case sensitive. are there any unusual characters in the path? did you put in the full path instead of the folder name by mistake? does the image show up on the list of images when you browse there? (perhaps you forgot to save)

@brakcounty use pip to install the mysql connector module for python

@brakcounty there is no way to “rescan” the images to rebuild the database entries, but here’s some python i used to list all images that exist as file folders but aren’t in the database

import mysql.connector
import os

mydb = mysql.connector.connect(
    host="localhost",
    database="fog",
    username="<valid database username>",
    password='<valid password>'
    )

def Diff(li1, li2):
    return (list(list(set(li1)-set(li2)) + list(set(li2)-set(li1))))

mycursor = mydb.cursor()
mycursor.execute("SELECT imagePath FROM images")
myresult = mycursor.fetchall()

dbimagelist = []

#print("########images in db")
for x in myresult:
    #print(x[0])
    dbimagelist.append(x[0])

#print("########directorys in /images")
#for x in os.listdir('/images'):
    #print (x)

#print("########diff")
for x in Diff(os.listdir('/images'),dbimagelist):
    print (x)

you may need to edit it for your environment

@nick the client communicates entirely with web based services on the fog server. in fact, if you know what to put in the url bar you can do most (maybe all, i can’t remember) of the call/response in a browser to see what gets returned when the client makes requests. which is something we do when troubleshooting.

@george1421 i meant to reply to this a long time ago, but here goes.

testing on deduping of those images has been done. they dedup quite well. the dedup changes affect zstd and pigz compressed images. pigz compressed images actually dedup better, but the the compression and performance are worse. it’s a tradeoff to be evaluated by the individual.

dedup is only possible with the newer version of partclone due to a rolling checksum integraed into the image format on earlier versions. the newer version lets us choose no checksum.

the compressed binary file is dedupable thanks to the --rsyncable flag on compression that is supported by both pigz and zstd.

like george said, any deduping would be the responsibility of the underlying filesystem or storage, not built into fog itself.

@fog_newb “FTP IMAGE SIZE” on the settings page