mutex lock on persistence updates

[Frank Fock]

Hi Dave,

Yes, you'll have to synchronize access to persistent data
of your MIB objects yourself. But using start_sync/end_sync
of your MibEntry derivate will not be suffient, since
persistent data is normally shared between objects in a MibGroup.
So, you will have to use the start_sync/end_sync (or
lock/unlock respectively) of the corresponding MibGroup.

Best regards,
Frank

Dave Mason wrote:
> Hi,
> I have an unusually simple question this time. :)  I use a subclass of 
> MibLeaf for most of my Mib objects, and I have a method which overrides 
> cleanup_set_request().  In there, I added some code that updates the 
> persistent storage for the MibGroup that the object belongs to.  I 
> havent stressed this agent too hard yet, but it occurred to me that if 
> multiple managers are doing updates at the same time on objects in the 
> same MibGroup, and both do persistence updates at the same time, could 
> the persistence file be corrupted somehow?  Should I add a 
> start_synch()/end_sync() around the persistence update?  In general, is 
> that true anytime I update persistence myself?  I didnt see that 
> happening in the toolkit anywhere.
> 
> Thanks,
> Dave
> 
> 
>