I am not a developer, but an IT manager. I noticed, from business standpoint, that once you have a large number of containers, with identical processes, LXC becomes unmanageable. I am the consumer of this technology. I am the "customer".
Yours Philip On Tue, May 20, 2014 at 11:16 PM, Fajar A. Nugraha <[email protected]> wrote: > On Wed, May 21, 2014 at 7:07 AM, CDR <[email protected]> wrote: >> Dear Friends >> I have 20+ containers with the same programs running. All of them are >> cpu-intensive. But one of them is eating way more CPU than the >> average. With "top" I have no idea which container "owns" that >> program. Perhaps we need a new "lxc-top" that would identify the >> process and the container, and maybe allow to sort by container-cpu or >> memory, or show cpu-container, memory-container, etc. > > > Sure. Submit patches to create the utility. > > You can start from output of lxc-ps and lxc-info. In your case, you'd > be most interested in CPU use. > > # lxc-ls --active > build > > # lxc-info -n build -S -H > CPU use: 18523054463 > BlkIO use: 2207232 > Memory use: 12247040 > Link: veth-build-0 > TX bytes: 8670717 > RX bytes: 227056 > Total bytes: 8897773 > > Depending on what you're used to, it might be easier to either write > something to extend snmp, or write a cacti plugin directly, and > monitor the output using cacti or > whatever-that-can-plot-raw-output-into-graph. > > -- > Fajar > _______________________________________________ > lxc-users mailing list > [email protected] > http://lists.linuxcontainers.org/listinfo/lxc-users _______________________________________________ lxc-users mailing list [email protected] http://lists.linuxcontainers.org/listinfo/lxc-users
