On Sat, Nov 28, 2020 at 10:25 PM Samuel Sieb <samuel@xxxxxxxx> wrote: > > On 11/28/20 4:13 PM, Chris Murphy wrote: > > On Fri, Nov 27, 2020 at 10:54 PM Sreyan Chakravarty <sreyan32@xxxxxxxxx> wrote: > > > >> Where exactly are you getting this from ? > >> > >> I have been using swap on LVM thin volume pretty much up around always, excluding this crash. > > > > It's possibly stale information, back from when the installer first > > added support for LVM thinp. I forget if it was someone on the > > installer or LVM team, but I can't find any reference. The installer, > > to this day, creates swap on a conventional "thick" provisioned LV, > > even when choosing the LVM thinp layout. > > I don't see why it would be any different on a thin LV as long as the > swap file is fully provisioned. It's still just a fixed set of blocks > on the hard drive. It's complicated. I'm suspicious because even in the corner of storage I mostly hang out in (Btrfsland) there are various logical to physical mapping issues. Hence all the limitations for swapfiles listed in man 5 btrfs. I know there isn't a shared/standard interface for finding the physical offset during early boot in order to restore a hibernation image, which is why we have an extra step to figure out that physical offset on Btrfs and add it as a boot parameter. Given my skepticism, I decided to start an upstream discussion. Unfortunately there's a problem with the redhat list archive not showing the thread. But it is on lore, which was just set up a couple days ago. https://lore.kernel.org/linux-lvm/608365664c2a18db4e756f524c0e76da@xxxxxxxxxx/T/#t Lore is awesome because it's searchable and there are quite a few other upstream linux lists stored there (including the file systems lists) Anyway, there are related issues to swap on the LVM list. e.g. it's not recommended to use it with LVM cache. https://lore.kernel.org/linux-lvm/57207A94.5040004@xxxxxxxxxx/ Switching back to hibernation, one of the additional issues there is the hibernation image must be made of contiguous blocks on the physical storage. And this isn't certain with either Btrfs or LVM thin. That's probably yet another thing we need a standard interface for when initially creating the logical to physical mapping. Or we need an updated hibernation image format that can be discontiguous. -- Chris Murphy _______________________________________________ users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx
