TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Bcachefs – A New COW Filesystem

292 点作者 jlpcsl大约 2 年前

18 条评论

londons_explore大约 2 年前
I would like to see filesystems benchmarked for robustness.<p>Specifically, robustness to everything around them not performing as required. For example, imagine an evil SSD which had a 1% chance of rolling a sector back to a previous version, a 1% chance of saying a write failed when it didn&#x27;t, a 1% chance of writing data to the wrong sector number, a 1% chance of flipping some bits in the written data, and a 1% chance of disconnecting and reconnecting a few seconds later.<p>Real SSD&#x27;s have bugs that make them do all of these things.<p>Given this evil SSD, I want to know how long the filesystem can keep going serving the users usecase.
评论 #35901547 未加载
评论 #35900987 未加载
评论 #35900961 未加载
评论 #35901610 未加载
评论 #35900847 未加载
评论 #35900975 未加载
评论 #35906031 未加载
评论 #35901006 未加载
评论 #35944346 未加载
评论 #35909835 未加载
评论 #35905099 未加载
评论 #35903418 未加载
voxadam大约 2 年前
You&#x27;re interested in more detailed information about bcachefs I highly recommend checking out <i>bcachefs: Principles of Operation</i>.[0]<p>Also, the original developer of bcachefs (as well as bcache), Kent Overstreet posts status updates from time to time on his Patreon page.[1]<p>[0] <a href="https:&#x2F;&#x2F;bcachefs.org&#x2F;bcachefs-principles-of-operation.pdf" rel="nofollow">https:&#x2F;&#x2F;bcachefs.org&#x2F;bcachefs-principles-of-operation.pdf</a><p>[1] <a href="https:&#x2F;&#x2F;www.patreon.com&#x2F;bcachefs" rel="nofollow">https:&#x2F;&#x2F;www.patreon.com&#x2F;bcachefs</a>
评论 #35907616 未加载
评论 #35909977 未加载
harvie大约 2 年前
What i really miss when compared to ZFS is ability to create datasets. I really like to use ZFS subvolumes for LXC containers. That way i can have separate sub-btree for each container with it&#x27;s own size limit without having to create partitions or LVs, format the filesystem and then resize everything when i need to grow the partition or even defragment fs before shrinking it. With ZFS i can easily give and take disk capacity to my containers without having to do any multi step operation that requires close attention to prevent accidental data loss.<p>Basicaly i just state what size i want that subtree to be and it happens without having to touch underlying block devices. Also i can change it anytime during runtime extremely easily. Eg.:<p>zfs set quota=42G tank&#x2F;vps&#x2F;my_vps<p>zfs set quota=32G tank&#x2F;vps&#x2F;my_vps<p>zfs set quota=23G tank&#x2F;vps&#x2F;my_other_vps<p>btrfs can kinda do this as well, but the commands are not as straighforward as in zfs.<p>update: My bad. bcachefs seems to have subvolumes now. there is also some quota support, but so far the documentation is bit lacking, so not yet sure how to use that and if that can be configured per dataset.
orra大约 2 年前
Oh! This is very exciting. Bcachefs could be the next gen filesystem that Linux needs[1].<p>Advantages over other filesystems:<p>* ext4 or xfs — these two don&#x27;t use ECC to protect your data, only the filesystem metadata<p>* zfs — zfs is technically great, but binary distribution of the zfs code is tricky, because the CDDL is GPL incompatible<p>* btrfs — btrfs still doesn&#x27;t have reliable RAID5<p>[1] It&#x27;s been in development for a number of years. It now being proposed for inclusion in the mainline kernel is a major milestone.
评论 #35899928 未加载
评论 #35900099 未加载
评论 #35899892 未加载
评论 #35902152 未加载
评论 #35902092 未加载
sporkle-feet大约 2 年前
The feature that caught my eye is the concept of having different targets.<p>A fast SSD can be set as the target for foreground writes, but that data will be transparently copied in the background to a &quot;background&quot; target, i.e. a large&#x2F;slow disk.<p>If this works, it will be awesome.
评论 #35900263 未加载
评论 #35901138 未加载
评论 #35909503 未加载
IAmLiterallyAB大约 2 年前
Here&#x27;s a link to the Bcachefs site <a href="https:&#x2F;&#x2F;bcachefs.org&#x2F;" rel="nofollow">https:&#x2F;&#x2F;bcachefs.org&#x2F;</a><p>I think it summarizes its features and strengths pretty well, and it has a lot of good technical information.
评论 #35901700 未加载
评论 #35901716 未加载
MisterTea大约 2 年前
Another File system I am interested in is GEFS - good enough fs (rather - &quot;great experimental file shredder&quot; until stable ;-). It&#x27;s based on B-epsilon trees, a data structure which wasn&#x27;t around when ZFS was designed. The idea is to build a ZFS like fs without the size and complexity of zfs. So far its plan 9 only and not production ready though there is a chance it could be ported to OpenBSD and a talk was given at NYC*BUG: <a href="https:&#x2F;&#x2F;www.nycbug.org&#x2F;index?action=view&amp;id=10688" rel="nofollow">https:&#x2F;&#x2F;www.nycbug.org&#x2F;index?action=view&amp;id=10688</a><p>Code: <a href="http:&#x2F;&#x2F;shithub.us&#x2F;ori&#x2F;gefs&#x2F;HEAD&#x2F;info.html" rel="nofollow">http:&#x2F;&#x2F;shithub.us&#x2F;ori&#x2F;gefs&#x2F;HEAD&#x2F;info.html</a>
baobrien大约 2 年前
huh, this is fun: <a href="https:&#x2F;&#x2F;lore.kernel.org&#x2F;lkml&#x2F;ZFrBEsjrfseCUzqV@moria.home.lan&#x2F;" rel="nofollow">https:&#x2F;&#x2F;lore.kernel.org&#x2F;lkml&#x2F;ZFrBEsjrfseCUzqV@moria.home.lan...</a><p>There&#x27;s a little x86-64 code generator in bcachefs to generate some sort of btree unpacking code.
评论 #35899937 未加载
评论 #35899938 未加载
评论 #35900718 未加载
eis大约 2 年前
I really hope Linux can get a modern FS into common usage (as in default FS for most distros). After more than a decade ZFS and BTRFS didn&#x27;t go anywhere. Something that&#x27;s just there as a default, is stable, performs decently (at least on ext4 level) and brings modern features like snapshots. Bcachefs seems to have a decent shot.<p>What I&#x27;d like to see even more though would be a switch from the existing posix based filesystem APIs to a transaction based system. It is way too complicated to do filesystem operations that are not prone to data corruption should there be any issues.
评论 #35900318 未加载
评论 #35903206 未加载
curt15大约 2 年前
For some reason VM and DB workloads are btrfs&#x27;s Achilles heel but ZFS seems to handle them pretty well (provided that a suitable recordsize is set). How do they perform on bcachefs?
评论 #35901742 未加载
brnt大约 2 年前
Erasure coding at the filesystem level? Finally!<p>I&#x27;ve not dared try bcachefs out though, I&#x27;m quite wary of data loss, even on my laptop. Does anyone have experience to share?
评论 #35901762 未加载
graderjs大约 2 年前
Is there an optimal filesystem or is it all just trade offs? And how far if we come since you know when we were first creating file systems plan nine or whatever to now like it’s as there been any sort of technological Leap like killer algorithm it’s really improve things?
评论 #35901444 未加载
评论 #35900662 未加载
throw0101b大约 2 年前
&gt; <i>These are RW btrfs-style snapshots</i><p>There&#x27;s a word for &#x27;RW snapshots&#x27;: clones. E.g.<p>* <a href="https:&#x2F;&#x2F;docs.netapp.com&#x2F;us-en&#x2F;ontap&#x2F;task_admin_clone_data.html" rel="nofollow">https:&#x2F;&#x2F;docs.netapp.com&#x2F;us-en&#x2F;ontap&#x2F;task_admin_clone_data.ht...</a><p>* <a href="http:&#x2F;&#x2F;doc.isilon.com&#x2F;onefs&#x2F;9.4.0&#x2F;help&#x2F;en-us&#x2F;ifs_t_clone_a_file_from_a_snapshot.html" rel="nofollow">http:&#x2F;&#x2F;doc.isilon.com&#x2F;onefs&#x2F;9.4.0&#x2F;help&#x2F;en-us&#x2F;ifs_t_clone_a_f...</a><p>* <a href="https:&#x2F;&#x2F;openzfs.github.io&#x2F;openzfs-docs&#x2F;man&#x2F;8&#x2F;zfs-clone.8.html" rel="nofollow">https:&#x2F;&#x2F;openzfs.github.io&#x2F;openzfs-docs&#x2F;man&#x2F;8&#x2F;zfs-clone.8.htm...</a><p>* <a href="http:&#x2F;&#x2F;www.voleg.info&#x2F;lvm2-clone-logical-volume.html" rel="nofollow">http:&#x2F;&#x2F;www.voleg.info&#x2F;lvm2-clone-logical-volume.html</a><p>In every other implementation I&#x27;ve come across the word &quot;snapshot&quot; is about read-only copies. I&#x27;m not sure why btrfs (and now bcachefs?) thinks it needs to muddy the nomenclature waters.
评论 #35901458 未加载
jacknews大约 2 年前
&quot;The COW filesystem for Linux that won&#x27;t eat your data&quot;<p>LOL, they know what the problem is at least. I will try it out on some old hard disks. The others (esp. looking at you btrfs) are not good at not losing your entire volumes when disks start to go bad.
renewiltord大约 2 年前
Is there a high-performance in-kernel FS that acts as a hierarchical cache that I can export over NFS? Presently I use `catfs` over `goofys` and then I export the `catfs` mount.
评论 #35909174 未加载
shmerl大约 2 年前
Cool to see it approaching upstreaming.<p>Does it support mountable subvolumes like btrfs?
layer8大约 2 年前
I parsed this as “BCA chefs” at first.
unixhero大约 2 年前
How does this compare with ZFS?