Posts made by tian

Hi,

I’ve encountered some problems maybe related to the queuing when deploying a group of computers. The groups have a size of 25 and 31 computers and the problem occurs in fog versions 1.5.10.1615 and also 1.5.10.1639.

The problem is that all the computers in the group are starting fine by WoL, booting up PXE an then successfully enter the queue to start imaging but none of them are starting imaging and display:
“There are open slots, but XX before me on this node (in line for yy:yy:yy)”

• XX is a higher number between 18 an 31 as far as I could see. Lower , single digit numbers seem not to be present.
• On the Dashboard the status is like: Free: 0, Queued: 31 (or 25), Active 0
• Under Tasks all the Imaging-Tasks are there and have the Status “Checked in” (Pause symbol)
• in the database table “tasks” there are no other Tasks with the taskStateID 1, 2 or 3 except the 31/25 that should be there
• The Max. Client Size of the (only) main node is set to 13 and has been tried to set to 8 or 10 but no computers started imaging

The “workaround” to get it started is to force 13 computers manually by clicking the lightning button in the active tasks. Then the PCs are starting to get the image. What I could monitor:

• after 13 computers are forced to image the queued comupters display: “No open slots, There are XX before me” with XX now lower numbers stating at around 15 but no lower or single digit numbers as far as I could see
• in the group of 25 computers the last 12 automatically begin to start
• in the group of 31 computers the status after 13 finish is:
  – Free: 0, Queued: 18, Active 0 but no new computers begin to start imaging.
  –After forcing another 3 computers to start, additionally 3 begin to start automatically and the status is: Free: 0, Queued: 12, Active 6.
  – Now the status is Free: 0, Queued: 0, Active 13 (witch seems to be normal)
  – after finishing all the tasks the status will be Free: 13, Queued: 0, Active 0 again

Single computer deplyoment works just fine.

Any Idea what could cause this problem? The version (1.5.10.something - I’ll try to find out the version number) we used before 1.5.10.1615 didn’t have this problem.

Thanks.

@AlexPDX since also the computer names are known it should be possible to add it to the database.
But it would not make much sense because when running a Hardware Inventory task or deploying a client (where also a Hardware Inventory task is run automatically) I think the data would be removed from the database.

To have the data in the fog database permanently I guess we have to wait untill FOS is able to get this data by itself and add it to the database.
For now to get the SSD model data without accessing each computer physically and have it in one place is enough for us to check for possible firmware updates, run the firmware updates and re-check if the updates were applied correctly.

we also have Dell computers, where the hard drive information is not in the Hardware Inventory (using fog 1.5.10.3 and AHCI in UEFI). To know the SSD models (e.g. for firmware updates) in the computers I made a snapin with a small batch file to get the information directly from Windows:

cd %~dp0 

net use x: <UNC-Path> /user:<domain>\<user> %1 /PERSISTENT:NO
wmic diskdrive get model,firmwarerevision,serialnumber >> X:\OptiPlex_HDD\%computername%.txt
net use X: /delete

The 3 values in <> needs to be customized. and the parameter %1 is the first snapin argument and the snapin arguments are set to hidden so the password is not in the local log files.

To combine the files I used:

for %f in (*.txt) do (type "%f" & echo %f) >> out.txt

and edit the final file manually to look better.

Maybe this helps a little bit. It might not be the best way to do this but we now have the model, firmware version and serial number for all the computers in one file.

@sebastian-roth thanks for the information. Since everything else is running fine there has been no need to try the dev-branch. It’s good to know that it will be included in the next release.

There seems so be a minor display issue with prime numbers involved since some results have infinite digits:


But it is not that important since everything else in 1.5.9 is working fine.

Hi matijn,

did you already try this?:
https://forums.fogproject.org/topic/10006/ubuntu-is-fog-s-enemy

We’re not on Ubuntu 18.04 yet and still running 16.04. But every time we update the OS by apt-get (doesn’t matter if upgrade or dist-upgrade) and it contains a SQL update the things described in the link above needs to be done.

If that solves your problem and you have some sort of automatic updates turned on you could think about to deactivate them and run them regularly by yourself.

If that doesn’t solve your problem maybe someone else has another idea.

Hi,

we just updated from fog 1.4.4 to fog 1.5.4. As far as we tested it looks fine but the progress bar when capturing/deploying an image behaves kinda weird:

At beginning all the information is squeezed in the little green status bar and after some progress there are more readable information. It’s the same in IE and Firefox.
(Capturing was tested on an older machine with higher zstd compression so the speed/time shown there is ok)

It is not that important but sometimes made it more easy to spot a computer with a slower network connection directly.

Thanks

@ravigon how about the values of igaPrimary (wonder why I wrote igaGroup …)?

What is the ID in “Storage Management” -> “All Storage Nodes” -> your storage node (normally “DefaultMember”) -> “Storage Group”? And are the other settings there correct?
Could you also check the Image Size log in “Fog Configuration” -> “Log Viewer” if there are any errors?

The Image Size Service seems to run every hour and check the image sizes (Fog 1.4.4):

...
[10-18-17 9:21:51 am] * Starting Image Size Service.
[10-18-17 9:21:51 am] * We are group ID: 1. We are group name: default
[10-18-17 9:21:51 am] * We are node ID: 1. We are node name: DefaultMember
[10-18-17 9:21:51 am] * Finding any images associated with this group as its primary group
...
[10-18-17 10:21:51 am] * Starting Image Size Service.
[10-18-17 10:21:51 am] * We are group ID: 1. We are group name: default
[10-18-17 10:21:51 am] * We are node ID: 1. We are node name: DefaultMember
[10-18-17 10:21:51 am] * Finding any images associated with this group as its primary group
...

But we once had a problem that its was not displaying the size for one or two images. The solution was that in the table “imageGroupAssoc” the value of igaGroup (?) was not “1” (I think it was 0).
After manually changing this value to 1 the next time the Image Size Service was running all of the images was showing the “Size on Server”. Dont’ know what caused this but we didn’t have this since then anymore.

So I think it’s worth checking the imageGroupAssoc table in the database and storage settings in FOG.

@george1421 this should be the correct solution
We’re also on Ubuntu 16.04 and the fog installer is failing at exactly the same point. After running the commands to alter the root user in mysql the installer runs fine.
I think this problem re-occurs even when just updating the system by apt-get sometimes - so I am running the mysql alter commands every time after apt-get to be on the safe side.

First of all thanks for this great project. We just switched our productive system for the students computer rooms from FOG 0.32 to version 1.4.4 and want to say something about what we experienced. Currently everything is working fine.

• The capturing and deploying is more fast and, at the same time, has a better compression (old: PIGZ_COMP = -6, new: Zstd 12)
• The display of the deploy/capture progress in the web interface is really great
• The new FOG client is reacting more fast after cloning and is saving some additional time
• A date filter for imaging log and snapin log is missing (but it will be added some time later - I asked that in an earlier topic)
• Snapin files once uploaded are now kept and are not deleted when uploading a new file to an existing snapin - (I just will cleanup the snapin directory from time to time manually - think I saw a SQL statement to list the files currently assigned to snapins somewhere in the forums to make it a little bit more easy)

We had a minor problem with the new client:

• The snapins (some batch scrips and sfx-archives) created for FOG 0.32 we could use without problems on 1.4.4 and the new client, except where we used reg.exe to change registry entries: without the parameter “reg:64” the registry entries were written to HKLM\Software\Wow6432Node\abc (In the batch file the path was “HKLM\Software\abc” -> without “Wow6432Node” !). Adding “reg:64” solved that and it also might be better to tell it explicitly to write in the 64bit or the 32bit part.

Something that is also different to FOG 0.32 is the queuing behavior (when unicast group deploying a room):

• In FOG 0.32 there always was was a fixed order which computer was cloned - in the new one its like first come first serve
  – when all computers are starting simultaneously there are more deploying at the same time than allowed in the storage node’s “max clients” option: e.g. 13 are allowed and it has been 14 to 16 for the first deploy batch of a room, the later computers normally will start when it is below 13 again but it can happen that 2 will start at the same time again (depending on the moment they try to check in I think) and it might be 12->14 again
• In FOG 0.32 the text for computers when waiting in the queue displayed the number of computer in before one (based on the fixed order?) - in the new FOG it is different
  – computers that started up with the other computers but didn’t get a free slot always display “0 before me”
  – computers that didn’t start up at beginning (WoL didn’t work, network cable was plugged out, …) are displaying the number of computers deploying plus the number currently queued as computers before one. (e.g. 14 computers deploying, 5 queued = “No open slots, There are 19 before me” for the 20th computer switched on)

The queue behavior describe above is not a problem for us - we just have to get used to it.

How are the load average on the machine? Without accessing the fog web interface I can see the following load average: 0,15, 0,14, 0,10 at the moment. When more computers are turned on / logged in the load is higher but still below 1 when I remember correctly.

While cloning (unicast) on the other side the average loads are very high at around 9 but everything still works fine.

@Tom-Elliott Thanks for your hard work - good to know it will be there in a future release.

In the Snapin Log the only date/time you can sort by is the crate date/time. This seems to make no sense for me since one snapin is created only once and even after updating/editing a snapin the create date/time wont change anymore.

I think the “Start/Checked in” and/or “Completed” date/time as columns you can sort by would be better. This way it would be more easy to verify if snapins were running fine.

Also a date range selection (like in FOG 0.32) for the “Snapin Log” and the “Imaging Log” would be nice if it is not too hard to implement.

Thanks

Here is the same message when updating 1.3.0 (revision 6050) to 1.3.1 (revision 6053) on Ubuntu 14.04 - it also says it can’t find the file, but the file is there in “/images/dev/postinitscripts/” and not “/images/dev/postdownloadscripts/”:

 ls -lisa /images/dev/postinitscripts/
insgesamt 12
3670017 4 drwxrwxrwx 2 root root 4096 Jan 16 09:18 .
 262145 4 drwxrwxrwx 5 root root 4096 Jan 16 09:18 ..
3670018 4 -rwxrwxrwx 1 root root  237 Jan 16 09:18 fog.postinit

The directory postdownloadscripts is here “/images/postdownloadscripts/” (without “/dev”) - but without the file “fog.postinit”.

ls -lisa /images/postdownloadscripts/
insgesamt 12
524289 4 drwxrwxrwx  2 root root 4096 Sep 23  2015 .
     2 4 drwxrwxrwx 10 root root 4096 Nov 11 14:42 ..
524290 4 -rwxrwxrwx  1 root root  233 Sep 23  2015 fog.postdownload

@Tom-Elliott Thanks for the solution. I’m using the new user.class.php now (renamed the not working one before) and the login with the previous user name (only letters and one “-” in the middle) is ok again.

(Since we don’t use the ldap integration I didn’t test the ldap.class.php)

Server

• FOG Version: 1.3.0 stable (revision 6050)
• OS: Ubuntu 14.04

Client

• Service Version: n/a
• OS: n/a

Description

I upgraded a server (from 1.3.0 RC25, revision 6020), where the user name contains a “-”. In RC25 the login worked correctly, in 1.3.0 final it won’t anymore and gives the message “invalid login” back.

After trying to find out why it doesn’t work for some time I simply removed the “-” from the user name by editing it in the database directly and it worked instantly with the edited username.

@Tom-Elliott Today we tested with Version 1.3.0-RC-1 (SVN Revision: 5935) and the “Chainload failed” message still appears with exit type “EXIT”.

But since the exit type “REFIND_EFI”works, we’re OK with that.

@Sebastian-Roth Thanks for the explanation.

@Tom-Elliott So far it is appearing on the Mac-System we maybe want to use fog with in the future.
I tried some more (now Version 8581) with the exit types and “REFIND_EFI” is working - instead of the “Chainloading failed” message a blue Refind screen appears for less then a second and OSX booting continues without the 10 seconds delay.

Currently we also use the trunk version for testing normal PCs for quite some time - but these don’t have EFI. I’ll ask my co-worker if he still has the older iMac (2007 Model I think) I used for testing PXE by the bless command a long time ago (Fog 1.20?) - because I don’t remember a “Chainloading failed” message back then.