[nycbug-talk] Advanced UNIX Basics Management

Bjorn Nelson o_sleep at belovedarctos.com
Thu Sep 24 08:03:19 EDT 2009


Isaac Levy wrote:
>      + top(1) (briefly, everyone knows top...)
>   
parallel to sar post event investigation, just dump top at 1 minute 
intervals to a file and copy it to nfs.  Very useful/worth that minimal 
perf impact.
>    - management tools
>      + kill(1), killall(1) (flags!)
>   
stop using kill -9 first
> - Memory
>    - stats/info facilities
>      + ps(1) (flags and some handy awk parsing)
>      + top(1) (briefly, everyone knows top...)
>      + swapinfo(8)
>    - management tools
>      - swapon(8), swapoff(8)
>   
linux just trust free + buffers + cache.  Get the devs started early on 
understanding linux never reclaims cache.

vmstat for overall perfs and gdb for core debugging (maybe they are 
already familiar with gdb).

-Bjorn



More information about the talk mailing list