Very slow cloning speed on specific model

[Middle]

@nrg The following is based on using Fog 1.5.7 master branch on CentOS.

Are you using the updated init_partclone.xz? Here’s the link:

https://drive.google.com/open?id=1u_HuN5NSpzb7YmQBAsrzDELteNmlWUWU

You need to copy this to /var/www/html/fog/service/ipxe and then run chown apache:apache init_partclone.xz to update file permissions.

We only have G6 laptops for imaging now, so I made this the default init under Fog Config > Fog Settings > TFTP Server > PXE BOOT IMAGE > enter init_partclone.xz. Alternatively, if you register hosts, you can add it to the host init section for just the G6 laptops.

You then have to edit the file /images/dev/postinitscripts/fog.postinit and add the line nvme set-feature -f 0x0c -v=0 /dev/nvme0

[bberret]

We have had some success with using the snponly.efi boot file instead of ipxe.efi. Also updating to fog version 1.5.7.100 might have helped as well. I did notice today HP came out with a newer bios verison for the HP 840 G6 model. Bios version 01.04.02 which seems to have a lot of bug fixes. I might give this a try tomorrow but I figured someone else might want to try it out before me and let me know how it goes.

[itsupport]

@darkxeno We have the same with Dell Ultra 7090
Updated the Kernal latest version of FOG stable
All our other Machines image in a few mins with SSD’s in them HDD’s 10 mins

These new ones take hrs which is not right
Doing one now and its like 200-300 mbps normally running at 1gig

Tested all our cables and like earlier other older machines like Latitude 5480 5490 5400
Desktops Opitplex 3020 3010 3050 3060 are all fine

[Sebastian Roth]

@itsupport This topic is fairly old and huge! No idea if this even talked about the DELL Ultra 7090 you seem to see an issue with. Please open your own topic and post details there (exact FOG and kernel versions and so on). Always good to post a link as reference to this topic as well.

[wuenschz]

@itsupport Hello, I was curious if a solution was ever discovered for fixing the slow imaging when working with Dell 7090 Ultras? We have recently purchased this model to refresh our labs, however we can’t get the image to push to the PCs. We are running FOG 1.5.9 and I just updated the Kernel to 5.10.34 TomElliot 64.

bzImage Version: 5.10.34
bzImage32 Version: 5.10.34

It seems to hang up for a while on “Restoring Partition Tables (GPT_)”

Any help would be greatly appreciated! Thank you!

[Sebastian Roth]

@wuenschz Have you read through the topic and tried things out yet? Let us know what you have tried so far and what didn’t work.

I would start by adding nvme set-feature -f 0x0c -v=0 /dev/nvme0 as a postinitscript - see: https://forums.fogproject.org/post/132360

[wuenschz]

@sebastian-roth Good morning Sebastian, I have been reading through the topic but have not tried anything beyond updating my FOG server’s Kernel to 5.10.34. I did see in the topic where it was suggested to add that parameter, however I believe it was stated that it was added in the 1.5.8 build of FOG. Should I still try adding it? We are using CentOS as well. If I do need to add that line, could you please clarify how I do that? I apologize but I’m not very good with Linux. Thank you for your help!

[wuenschz]

@sebastian-roth Hello again, I just wanted to update you on this issue. I have launched the task in debug mode and entered nvme set-feature -f 0x0c -v=0 /dev/nvme0. I am now able to get past the “Restoring partition Tables (GPT)” portion of the cloning, however it still gets hung up at the partclone screen. I did look at the partition information and it said that “Disk /dev/nvme0n1: 256GB” Do I have to change the command to reflect nvme0n1 instead of the /dev/nvme0?

Also, I saw that it says “Write caching not supported” but does support the drive and sees the drive.

I have also tried adding the nvme_core.default_ps_max_latency_us=0 kernal arguments to to the host in FOG and that did not fix anything.

[Sebastian Roth]

@wuenschz said in Very slow cloning speed on specific model:

Do I have to change the command to reflect nvme0n1 instead of the /dev/nvme0?

I don’t have any of these disks myself to test, so I am not sure. Give it a try with /dev/nvme0n1 and let us know.

[wuenschz]

@sebastian-roth Good morning,
Yesterday I tried changing the command to nvme0n1, however that did not make a difference. Today I am going to try rolling the Kernel back a version and see if that makes a difference. It is weird that it is transmitting data, just extremely slowly. The 7090 Ultras are transferring at 70-200MB per minute and our other devices transfer anywhere from 4-10GB per minute.

[herman]

@Middle said in Very slow cloning speed on specific model:

init_partclone.xz

Link down of Drive Google…!!!
Have another place or link?

[Sebastian Roth]

@herman said in Very slow cloning speed on specific model:

Link down of Drive Google…!!!

Well, this link is more than two years old and was probably meant as a temporary one…

What exactly so you hope to get from an old init.xz? I suggest you open a fresh new topic here in the forums and describe the exact issue you have, mention the FOG version and OS version used.