|
@mgattozzi | |||||
|
It has been 0 days since I showed coworkers the cursed Instant::now() code from Rust stdlib pic.twitter.com/V5Rdzw0kDi
|
||||||
|
||||||
|
to kubectl or not to kubectl that is the question
@mgattozzi
|
29. sij |
|
|
||
|
Geoffroy Couprie
@gcouprie
|
29. sij |
|
Oh no I had completely forgotten about this and it is the most cursed thing. Can't even trust time
|
||
|
|
||
|
to kubectl or not to kubectl that is the question
@mgattozzi
|
29. sij |
|
Trust no one, not even yourself
|
||
|
|
||
|
iximeow
@iximeow
|
29. sij |
|
following up on the relevant linux bugs is fun: twitter.com/iximeow/status…
|
||
|
|
||
|
to kubectl or not to kubectl that is the question
@mgattozzi
|
29. sij |
|
Excuse me
|
||
|
|
||
|
NickH
@NickHansen600
|
29. sij |
|
@lukechampine you'll love this
|
||
|
|
||
|
Luke Champine
@lukechampine
|
29. sij |
|
Oof. I wonder how Go handles it...
|
||
|
|
||
|
Tensibai
@Tensibai
|
29. sij |
|
May be dumb, BUT could it be related to systems using ntp under virtualization? The problem was widely known at a time before VMware add the sync clock to host as ntp was correcting too much when the cpu got the missing ticks from suspension
|
||
|
|
||
|
to kubectl or not to kubectl that is the question
@mgattozzi
|
29. sij |
|
It's not ntp. It has to do with clocks in things like the CPU and how the os gets them and interprets it.
|
||
|
|
||
|
Wesley Wiser
@wesleywiser
|
29. sij |
|
The prior version of this code wasn't so comprehensive and I broke some people's rustc by relying on this being monotonic 😬
github.com/rust-lang/rust…
|
||
|
|
||
|
Wesley Wiser
@wesleywiser
|
29. sij |
|
So if you want proof this happens not just in theory, there's all the duplicate issues.
|
||
|
|
||