Restarting the MemcacheD Cluster

[Philip Neustrom]

Why would a rolling restart cause cache corruption?  You mentioned
that you had cache corruption to begin with (in your application?) 
Maybe this is why you see it on a rolling restart?

 --Philip Neustrom

On 4/14/06, timeless <time at digg.com> wrote:
> Suppose somehow we've decided there's "bad data" in the MemcacheD
> cluster, and don't really know where it is. It seems the safest course
> of action would be to flush_all across the cluster.
>
> Here's what I came up with for doing it:
>
> for i in 1 2 3 4 5 ; do echo "flush_all" | netcat -q 1 memcached.$i
> 11211 ; sleep 15 ; done
>
> Does this seem reasonable to everyone? Is there a more elegent (or
> proper) way of doing this? (Doing a rolling restart of the memcached
> daemons across my cluster appears to be the WRONG way to do it, and
> seems it would cause cache corruption).
>
> --
> timeless
>