BitTorrent restore question - peers count, restore issues

[Solomon Krusader]

Hi Steven
Thank You for the great software

We were using Clonezilla lite-server to restore systems via PXE from image using multicast.

Recently we stopped making images, and switched to restoring systems from raw device - RAM Disk option, and stopped using multicast in favor of BitTorrent.

Multicast restore is pretty easy to comprehend - all or nothing - all systems should go hand in hand packet by packet, and if one fails ( it does happen from time to time ) restore stops for all and cannot proceed.
When using BitTorrent on other hand - You can reboot stuck system in the middle of restore and it will start over.. that's it.

All systems connected with 1Gbps links.

Questions/Issues:
1. We observed maximum 3 peers and there is no place where it can be configured in TUI.
2. Can something be done that clients will stay up after completing restore to 100% and continue to upload to restoring systems?
3. Sometimes we observed 0 peers although we clearly see that there are peers, that could upload to restoring system.
4. Sometimes speed is dropped to extremely low values ~ 0.2GB/min with 3 peers available.
5. Restore stuck at some point on couple of systems and does not move forward even after starting over but restore from raw device as remote-source/remote-destination works fine.
6. Can You point me to a doc or just explain what does netboot - use existing dhcp option mean - we have Active Directory DHCP server and we do not want to interfere with it in any way - so we choose start new dhcp instance in Clonezilla lite-server and disconnect whole classroom switch from the network (another reason is multicast in a network that does not have IGMP support). As long as BitTorrent technology does not use multicast and there is no fear to put down the network on its knees, we thought maybe we can choose use existing dhcp. Is it OK in AD environment to choose this option?

Thank You

UPD:
Unfortunately today reimaging of classroom via bittorrent ended up stuck on all classroom (20 clients).
Switched to multicast from raw device - speed ~ 2GB/min that is strange 1Gib network will give You approx. 6.5GB/min at least.

[Steven Shiau]

Which version of Clonezilla live did you use?
Have you tried the latest stable Clonezilla live, i.e., 3.1.3-16 or even testing one, i.e., >= 3.1.3-22?
https://clonezilla.org/downloads.php

Steven

[zbcgn]

Hey there,

sorry for bringing this up again, but i couldn't find any entries regarding these issues.

has this been resolved? I'd imagine 1. and 2. would be very helpful to improve deploy speeds.

1. We observed maximum 3 peers and there is no place where it can be configured in TUI.
2. Can something be done that clients will stay up after completing restore to 100% and continue to upload to restoring systems?

Especially 2. heavily influences speeds at the end of the process, when more and more client disconnect and dont act as seeders anymore. I tried the parameter "-true" for clients to "do nothing" after the process, but they still exit the EZIO interface. Is there a way to keep them in bittorrent seeder mode?

I'm working with 3_2_1-9-AMD

thank you

[Steven Shiau]

Please give testing Clonezilla live >= 3.2.2-1 or 20250507-* a try:
https://clonezilla.org/downloads.php
At the moment we assigned 60 secs (was 15 secs) for the leecher to wait and keep uploading.
We will add an option in the later release to let you assign the timeout time.

Steven

[zbcgn]

wow thank you for your amazing work!

[Steven Shiau]

We have assigned 3 ezio (BT) related parameters in Clonezilla live >= 3.2.2-5 or 20250512-*.
For more info, please check the changelog.txt:
https://clonezilla.org/downloads/testing/changelog.php

[zbcgn]

i have tested 3.2.2-5 and both changes to ezio max seed and max upload already have a very positive effect!

thanks a lot!

[Steven Shiau]

Cool! If you have the results between the version 3.2.1-* and 3.2.2-5, please share that, like the time to deploy the system, the nodes you have deployed, and the 3 parameters you used in 3.2.2-5.
Thanks.

[zbcgn]

From what I have tried, the following configuration seems to be the sweetspot for my purposes:

ezio_seed_max_connect="8" and ezio_seed_max_upload="4"

Regarding deploy times, i'd rather share transmission speeds. Of course, I'm hoping for maximum speeds limited by the 1 Gbit network switch(es) I'm using. The latest deployments (20-30 simultaneous PCs) had up- and download speeds of ~1 Gbit/s, thus happily operating at full capacity.

Hope this helps. Thanks for your continuous work!

[Steven Shiau]

Cool! Thanks for sharing that.

[Steven Shiau]

BTW, just curious. The machines you deployed use hard drive or SSD? If you do not mind, please share its brand and model.
Thank you very much.