#11670 NORM 1.75-so: olpc-update-query does not account for missing timestamp files

Zarro Boogs per Child bugtracker at laptop.org
Mon Mar 5 20:54:26 EST 2012


#11670: olpc-update-query does not account for missing timestamp files
---------------------------------+------------------------------------------
           Reporter:  greenfeld  |       Owner:  martin.langhoff                  
               Type:  defect     |      Status:  closed                           
           Priority:  normal     |   Milestone:  1.75-software                    
          Component:  distro     |     Version:  Development build as of this date
         Resolution:  invalid    |    Keywords:                                   
        Next_action:  no action  |    Verified:  0                                
Deployment_affected:             |   Blockedby:                                   
           Blocking:             |  
---------------------------------+------------------------------------------
Changes (by greenfeld):

  * status:  new => closed
  * next_action:  diagnose => no action
  * resolution:  => invalid


Comment:

 I convinced myself with a spreadsheet and some playing around that the
 current scheme will work except in the case where the last attempt flag
 has a future date.

 olpc-update-query does set the flag files; it just was doing so in a
 manner which confused me.

 Minor possible bugs if I'm reading the code correctly:
  * #11669
  * If lease.sig is missing  we do not force a run on secured XOs.  This
 should not happen under normal conditions.  {May not be desirable if a
 developer key is active.}
  * If the last attempt timestamp is in the future, the probability of us
 contacting the server becomes zero.
  * The last attempt flag is set before a query could potentially reset the
 RTC clock.  The last query flag is set afterwards.
  * The query timestamp is updated regardless of if we successfully
 contacted the OATS/XS servers or not.  A few comments in the code refer to
 this as the "last successful query," which is what made me believe the
 randomization code was not functioning in the first place.

-- 
Ticket URL: <http://dev.laptop.org/ticket/11670#comment:3>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system


More information about the Bugs mailing list