[CALUG] Strange behaviour

[John Szakmeister]

On Tue, Jun 4, 2013 at 8:09 AM, Rajiv Gunja <opn.src.rocks at gmail.com> wrote:
> This is not on Linux, but on Solaris/Unix.
>
> I have a few servers on which I am running a few commands via cron.
> Examples commands: date, uptime, sar, netstat, dladm, prstat, ping.
>
> Here is the command sequence I am having the issue:
>
> ldavg=`uptime | cut -d":" -f4`
>
> DDATE1=`date +'%Y.%d.%h.%H.%M.%S'`
>
> echo "$DDATE1 >>> $ldavg" >> /var/adm/save/2ping/load.avg.log 2>&1
>
> In the log file, on some servers, once every hour, it will miss output of
> the variable $ldavg. So the entried will look like this :

That is strange.  I initially thought maybe you were hitting a time
where perhaps the time didn't have the extra colon for the seconds if
you hit right on the minute boundary when executing uptime.

> 2013.04.Jun.07.23.24 >>> 1.12, 1.23, 1.25
>
> 2013.04.Jun.07.24.29 >>> 1.03, 1.18, 1.23
>
> 2013.04.Jun.07.25.35 >>> 1.10, 1.18, 1.22
>
> 2013.04.Jun.07.26.40 >>> 1.45, 1.30, 1.26
>
> 2013.04.Jun.07.27.45 >>>

I don't think this can be due to missing the output (I hope that
couldn't happen!).  Rather, since the line is there at all, I think
it's more like the input string changed in a slightly unexpected way,
so you're asking for a field that doesn't exist, which will come back
empty.

>
> 2013.04.Jun.07.28.50 >>> 1.10, 1.23, 1.24
>
> 2013.04.Jun.07.29.55 >>> 1.08, 1.20, 1.23
>
> 2013.04.Jun.07.31.01 >>> 1.66, 1.29, 1.26
>
> 2013.04.Jun.07.32.06 >>> 1.27, 1.25, 1.25
>
>
> I have no explanation as to why this would miss and that too at exactly that
> time every hour. My script runs these commands once a minute, every hour.

Perhaps a different tactic will work with the varying number of colons:

    uptime | sed -e "s/^.*: //"

Just eat everything until the last colon and space, and what's left
will be your averages.

-John