[Linux-cluster] gfs atime perf question (SOLVED)

[FM]

I use md5sum now

FM wrote:
> Hello,
> I just installed a 4 nodes webfarm (using direct routing).
> a cron on each server check a file ($netconfig on GFS) like this :
>
> if ! find $netconfig -anewer $netconfig | grep -q $netconfig ; then
>     echo "cluster node $node : $netconfig has changed, reconfiguring
> network ..."
>     "$netsetup"
> fi
>
> So when I update the file, every server run $netsup. this is my way to
> configure IP and arptables.
>
> But as I know now, the atime is updated every hour. So my first option
> would be to tune this to 60 instead of 3600. But I suppose that I will
> have a perf issue ?? Do you have tips to handle that kind of scenario :
> watch a file for modification ?
>
> Thanks !
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>