TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Two Objects Not Namespaced by the Linux Kernel (2017)

169 pointsby setraover 6 years ago

9 comments

habermanover 6 years ago
&gt; The current set of namespaces in the kernel are: mount, pid, uts, ipc, net, user, and cgroup. [...] [Time is] not namespaced. [...] The kernel keyring is another item not namespaced.<p>I&#x27;ve always argued that &quot;everything is a file&quot; is an exaggeration. These moments make the extent of that exaggeration clear.<p>If everything truly was a file, the only thing you would need to namespace is the filesystem. But in reality there are a lot of other kernel objects that are not files at all.
评论 #18349857 未加载
评论 #18350392 未加载
评论 #18349907 未加载
评论 #18350488 未加载
评论 #18349790 未加载
评论 #18351356 未加载
评论 #18351825 未加载
评论 #18349797 未加载
评论 #18353400 未加载
wmfover 6 years ago
Since this was written a time namespace was proposed: <a href="https:&#x2F;&#x2F;www.phoronix.com&#x2F;scan.php?page=news_item&amp;px=Linux-Time-Namespace-RFC" rel="nofollow">https:&#x2F;&#x2F;www.phoronix.com&#x2F;scan.php?page=news_item&amp;px=Linux-Ti...</a>
评论 #18350511 未加载
derefrover 6 years ago
I wonder whether namespacing time would also result in those namespaces being able to have separate &quot;clocks&quot; (time backends? time schedulers?) that progress at different rates, or for different reasons.<p>Being able to put a process into a time namespace with a deterministic &quot;clock&quot; would obviate a large benefit of <a href="http:&#x2F;&#x2F;www.zerovm.org&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.zerovm.org&#x2F;</a>.<p>Also, having &quot;clock slew&quot; be a matter of perspective—with processes that <i>can</i> handle leap seconds seeing them happen instantaneously; and processes that <i>can&#x27;t</i> handle leap-seconds, seeing slewed time—would be nice. Then you could have different system facilities that care about <i>monotonic</i> time, vs. <i>synced to calendar</i> time, vs. <i>one second per second</i> time, all having that kind of time available to them as &quot;the time&quot;, rather than through different APIs.
评论 #18349773 未加载
评论 #18349579 未加载
评论 #18351925 未加载
评论 #18349893 未加载
评论 #18352114 未加载
theamkover 6 years ago
I personally miss core pattern namespacing. I would love to give some of my containers a custom coredump handler, but this is impossible.<p>And in general, a sysctls settings namespace would be really useful. Sure, sometimes it makes no sense to namespace a setting, but net.ipv4.tcp_congestion_control for example? I&#x27;d love to be able to change it without modifying the code.
vxNsrover 6 years ago
meta: This is from 2017,<p>Super interesting though, the keyring thing especially seems to have broader implications...
tyingqover 6 years ago
Syslog seems to be on the proposal list as well.
lalaithionover 6 years ago
Why is this the case? No one has bothered to do it? It would break backwards compatibility? Linus thinks it&#x27;s a bad idea?
评论 #18349405 未加载
评论 #18351532 未加载
评论 #18349426 未加载
Sharlinover 6 years ago
I’m not sure that people who think ”containers are just like VMs” should have any business working with containers.
timeattackover 6 years ago
You can&#x27;t change time in container, but it&#x27;s possible to change timezone files.<p>With generating fake timezones it is possible to change time in container.
评论 #18351929 未加载