Re: [HTCondor-users] convert from 'date +"%H-%M %m-%d-%Y"' to seconds from 1970

[John M Knoeller <johnkn@xxxxxxxxxxx>]

Dots *are* allowed in macro names, but not in  ClassAd Attribute names,  and this configuration is using Atlas.Seconds for both types of name which is not going to work. 

On the other hand.  This seems like mixing up namespaces to me

> AtlasSeconds = 1588874264
> STARTD_ATTRS = $(STARTD_ATTRS), AtlasSeconds, FITS_IN_MAINTENANCE
> FITS_IN_MAINTENANCE = (time() + TARGET.RequestRuntime) < $(AtlasSeconds)
> START = $(START:True) && (FITS_IN_MAINTENANCE)

$() expansion of AtlasSeconds here means there is no need to insert it into the STARTD attrs, because it
is being substituted when the configuration is used.  meaning that FITS_IN_MAINTENANCE is really this

FITS_IN_MAINTENANCE = (time() + TARGET.RequestRuntime) < 1588874264

So I think you either want this

   AtlasSeconds = 1588874264
   FITS_IN_MAINTENANCE = (time() + TARGET.RequestRuntime) < $(AtlasSeconds)
   START = $(START:True) && (FITS_IN_MAINTENANCE)

Or this

  AtlasSeconds = 1588874264
  STARTD_ATTRS = $(STARTD_ATTRS), AtlasSeconds, FITS_IN_MAINTENANCE
  FITS_IN_MAINTENANCE = (time() + TARGET.RequestRuntime) < AtlasSeconds
  START = $(START:True) && (FITS_IN_MAINTENANCE)

Also AtlastSeconds is a fixed point in time that is being inserted into the configuration, That makes sense
if the STARTD itself has a very limited lifetime (like a glidein).  But it makes no sense for a long-lived startd

Is this for a glidein?
 
-tj


-----Original Message-----
From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> On Behalf Of Carsten Aulbert
Sent: Wednesday, May 6, 2020 6:05 AM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>; Beyer, Christoph <christoph.beyer@xxxxxxx>
Subject: Re: [HTCondor-users] convert from 'date +"%H-%M %m-%d-%Y"' to seconds from 1970

Hi

On 5/6/20 12:56 PM, Beyer, Christoph wrote:
> FITS_IN_MAINTENANCE = (time() + TARGET.RequestRuntime) < Atlas.Seconds

Just a shot into the dark: $(Atlas.Seconds)?

(Are dots allowed in variable/macronames?)

Cheers

Carsten


-- 
Dr. Carsten Aulbert, Max Planck Institute for Gravitational Physics,
CallinstraÃe 38, 30167 Hannover, Germany
Phone: +49 511 762 17185