Interesting failure mode I have seen a few times now: App runs in Docker on laptop, OOMs in #kubernetes. Turns out to be ...
-
-
Replying to @thockin
... code that scales usage per-core or per-GB. It detects many cores (more than laptop) via /proc/cpuinfo, and tries to use them, but ...
4 replies 0 retweets 8 likes -
Replying to @thockin
... the container is limited to less-than-all the cores on the machine, so it blows up. Lesson: ...
1 reply 0 retweets 5 likes -
Replying to @thockin
... how to probe "available" resources is necessarily different from inside containers. This is under-recognized and under-developed ...
4 replies 6 retweets 18 likes -
Replying to @thockin
... We need new tools & techniques to learn what you can ACTUALLY use in your container, rather than what the machine has installed.
4 replies 0 retweets 18 likes -
Replying to @thockin
And unfortunately not necessarily that easy unless the container is setup in a particular way (e.g. read access to cgroup)
2 replies 0 retweets 3 likes -
Replying to @cpuguy83
True. Without cgroup namespaces, having them mounted inside the container is an info leak at best, and possibly a security/isolation problem
1 reply 0 retweets 2 likes -
One thing
#kubernetes does is let you map resource limits to env vars and virtual files.2 replies 0 retweets 3 likes -
Original issue: https://issues.k8s.io/9473
1 reply 0 retweets 0 likes
Original original issue: https://github.com/kubernetes/kubernetes/issues/386#issuecomment-57265558 …
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.