909546 - Support for new_crashes with Monitor with a configuration that uses a single filesystem (no crashmover)

Status: RESOLVED INVALID

(Socorro :: Backend, task)

Description

[Selena Deckelmann :selenamarie :selena]

A user is trying to set up Socorro using a single filesystem. 

I believe the correct CrashStorage module to use is FSLegacyDatedRadixFilesystemStorage. However, this class doesn't implement "new_crashes". 

Two questions: 

* Should we implement new_crashes() in this class? 
* Is there an alternative filesystem storage we could use instead?

Comment 1

[[DEACTIVATED] Adrian Gaudebert]

I wonder if that's related to the problems I'm having with bug 908798? I suppose so, if this bug says "Socorro doesn't work with FSLegacyDatedRadixFilesystemStorage".

Comment 2

[K Lars Lohn [:lars] [:klohn]]

"...using a single filesystem"

  1) does that mean no HBase and the file system is to be permanent storage for all crashes?  
  2) will the new system be using throttling?

since it is a new system, it doesn't need to be compatible with our traditional (legacy) file storage structure.  The class FSDatedRadixTreeStorage should work just fine.  If throttling is to be used, then the class socorro.crashstorage_base.PrimaryDeferredStorage should be used with FSDatedRadixTreeStorage as primary and FSRadixTreeStorage as deferred.  

> I believe the correct CrashStorage module to use is FSLegacyDatedRadixFilesystemStorage. 

This is the class that is completely backwards compatible with our old (legacy) file system.  This is the class that we are using on our collectors right now.

> However, this class doesn't implement "new_crashes".

Its base class FSDatedRadixTreeStorage does implement "new_crashes".  "new_crashes" is called by our crashmovers on the class FSLegacyDatedRadixFilesystemStorage and it works just fine.

Comment 3

[Lonnen :lonnen]

monitor is no longer a component of socorro