| kata-irc-bot | <eric.ernst> :thread: maximum hot-pluggable memory | 16:07 |
|---|---|---|
| kata-irc-bot | <eric.ernst> It takes a certain amount of memory to online hotplugged memory, resulting in keeping a relatively higher default amount of memory when booting the guest. | 16:08 |
| kata-irc-bot | <eric.ernst> I'm re-reading my notes, throwing here for folks to recall. With a 2048 MB guest, you can only hotplug ~119 GB. | 16:14 |
| kata-irc-bot | <fidencio> The maximum hot-pluggable memory should be the maximum amount of free memory you have on the host side, no? I'm assuming you have a host with 128GB of RAM, and you're using QEMU, right? | 19:33 |
| kata-irc-bot | <eric.ernst> Nah, you need a certain amount of memory to online memory itself. | 19:36 |
| kata-irc-bot | <eric.ernst> Ie, it needs to allocate data structures for the new memory. | 19:36 |
| kata-irc-bot | <fidencio> Aha, I see. I will just take the backseat and learn from this thread. :slightly_smiling_face: | 19:37 |
| kata-irc-bot | <fidencio> For my future self to check later on: https://www.kernel.org/doc/html/latest/admin-guide/mm/memory-hotplug.html https://github.com/qemu/qemu/blob/master/docs/memory-hotplug.txt | 19:40 |
| kata-irc-bot | <eric.ernst> It's the guest Keene behavior that's the limitation here (acpi memory device handling). | 19:48 |
| kata-irc-bot | <eric.ernst> Though, I suppose we could try to break large memory additions into chunks to avoid shooting ourselves in the foot. | 19:49 |
| kata-irc-bot | <eric.ernst> Ie, only online / hotplug 100 GB at a time | 19:50 |
| *** xolcolor4 is now known as xolcolor | 23:56 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!