#9001 NORM Not Tri: Experiment to understand rpm/yum resiliency to sudden loss of power

[Zarro Boogs per Child]

#9001: Experiment to understand rpm/yum resiliency to sudden loss of power
-----------------------------+----------------------------------------------
 Reporter:  martin.langhoff  |                 Owner:  martin.langhoff
     Type:  task             |                Status:  new            
 Priority:  normal           |             Milestone:  Not Triaged    
Component:  school server    |               Version:  not specified  
 Keywords:                   |           Next_action:  never set      
 Verified:  0                |   Deployment_affected:                 
Blockedby:                   |              Blocking:                 
-----------------------------+----------------------------------------------
 The core mission is to understand how good rpm and yum are at recovering
 after a bad power loss (on our default fs, in this case ext3).

 The basic idea is to break yum/rpm install process by yanking the power
 away from the machine at different points of the installation, and seeing
 how the overall process recovers. And harassing or contratulating Seth and
 the rpm folks with what we learn :-)

 One idea: Perhaps some of it can be automated with qemu and a
 script that calls yum/rpm via ssh. Qemu has slow IO so you can get
 more granularity...

-- 
Ticket URL: <http://dev.laptop.org/ticket/9001>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system