[clue] tee in cronjob

[Quentin Hartman]

oh, and as far as my implied question of "what if I want more email?", now
that I have the context of the while being in a subshell, the most
straightforward approach seems to be to initiate the while loop with a
variable, and split the reporting/logging of the error from the detection
of the error. That way, you can make the loop exit on error detection, and
then report and log the error outside the while without unusual for loops
or weird redirection.

Q


On Mon, May 19, 2014 at 5:32 PM, Quentin Hartman <qhartman at gmail.com> wrote:

> I hadn't considered that the while loop opens a subshell, that's almost
> certainly it.
>
> Thanks!
>
> Q
>
>
> On Mon, May 19, 2014 at 5:20 PM, Mike Nolte <obiwanmikenolte at gmail.com>wrote:
>
>> QH,
>>
>> I'm a little unclear on the specifics, but a "while" loop opens a
>> subshell, which could be why it's being buffered.  You could use a for
>> loop, use redirection instead of tee (>>), or something like that.
>>
>> ~Monkeypenny
>>
>>
>> On Mon, May 19, 2014 at 5:09 PM, Quentin Hartman <qhartman at gmail.com>wrote:
>>
>>> I have a script that looks roughly like:
>>>
>>> now=`date +%F`
>>> logfile = log-$now.log
>>> while [ 1 -eq 1 ]
>>> do
>>>  things
>>>   if [ $? -ne 0 ]
>>>   then
>>>    echo "Oh noes!" | tee -a $logfile
>>>   fi
>>> done
>>>
>>> And it's run via cron, and everyday at midnight cron kills it and
>>> relaunches it. I do this to make sure it gets restarted if something goes
>>> Terribly Wrong, and also to rotate the log output into the next day's file.
>>> I know I could setup a log rotate entry for it, but I wanted this to be
>>> self-contained.
>>>
>>> Interestingly, all of the "tee" output seems to be buffered until cron
>>> kills the script or the script dies for some other reason. This means I
>>> only get one email each day from cron that rolls up all the various errors
>>> from that day in a single message. This is actually desirable behavior in
>>> this case, but I don't understand _why_ it's happening. What if I wanted to
>>> be email immediately when an error was detected and also log the error? I'm
>>> not seeing anything in my usual reference materials that would explain this
>>> behavior.
>>>
>>> Any ideas? Is there a better way to to execute this sort of thing?
>>>
>>> QH
>>>
>>> _______________________________________________
>>> clue mailing list: clue at cluedenver.org
>>> For information, account preferences, or to unsubscribe see:
>>> http://cluedenver.org/mailman/listinfo/clue
>>>
>>
>>
>> _______________________________________________
>> clue mailing list: clue at cluedenver.org
>> For information, account preferences, or to unsubscribe see:
>> http://cluedenver.org/mailman/listinfo/clue
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://cluedenver.org/pipermail/clue/attachments/20140519/a768bfb3/attachment.html