[Devel] Re: [lxc-devel] Containerized syslog

[Enrico Weigelt]

* Daniel Lezcano <daniel.lezcano at free.fr> schrieb:

> I thing a consensus was not reach. The big deal with syslog is netfilter 
> logs in an interrupt context where it is difficult to find the right log 
> buffer ring as we are not in the process context making possible to 
> identify the namespace.

That's one of the BIG problems in the Linux kernel, we've recently
stumbled with in another project (artificial disk bw limits): 
far too much happening within interrupt context. 

Interrupts were originally as _short time_ step-ins for time critical 
signaling. What's happening here is an abuse of that concept, w/ really
bad implications (eg. arbitrary users can create loads to bring down 
the whole system on extreme load). A clean and more robust solution 
IMHO could be that the interrupt handler only loads the received l1 
frames and enqueue them for an kernel thread (or multiple ones) - an 
containerized setup then would have separate queues and kthreads on 
per-container basis.


cu
-- 
---------------------------------------------------------------------
 Enrico Weigelt    ==   metux IT service - http://www.metux.de/
---------------------------------------------------------------------
 Please visit the OpenSource QM Taskforce:
 	http://wiki.metux.de/public/OpenSource_QM_Taskforce
 Patches / Fixes for a lot dozens of packages in dozens of versions:
	http://patches.metux.de/
---------------------------------------------------------------------
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers