[Insight-developers] [Cdash] CDash back online

[Bill Hoffman]

Bill Lorensen wrote:
> Julien,
> 
> Any luck with the submissions for last night's missing builds for
> ITK's dashboard? We have had some changes to pervasive classes and I'd
> hate to lose yesterday's submissions.
> 
They should be still in Dart1 on public.  It is a matter of moving over 
the XML files.  I can try to do this.

> In the future, will this be a recurring problem if the database goes
> down or does not recover automatically from system failures? Is there
> anyway to have the database repair itself when the host system
> recovers after a power failure or crash? Or will manual intervention
> always be required?
> 

In this case, the whole system crashed.  It was not just a DB issue. I 
came in in the kernal had seg faulted somewhere in the file system code 
(there was a call stack on the screen). I think that the file system was 
corrupted from the previous power outages without UPS.  However, the OS 
mounted the unstable FS.  This error would have been the same for Dart1, 
Dart2 or CDash, and would have required manual intervention.  I think we 
have it under control at this point.  We have done the following:

1. put the system on UPS
2. rebuilt the File System

In the future, if the system should go down for some reason we will 
always make sure the FS is checked and repaired.  In summary, I think we 
are getting things worked out with this machine.  We are making progress 
with each crash towards avoiding future crashs.  :)


-Bill