[ogfs-dev]memexpd's role

["Cahill, Ben M" <ben.m.cahill@intel.com>]

Hi all,

I was looking through the "ogfs-redesign" doc, along with some of the memexp and memexpd code ... there seem to be some discrepancies, but I wanted to run them by the list before making changes to the redesign doc.  

It looks to me that memexpd is not a "lock server", but merely a "storage server".   The memexp lock module can use either DMEP (SCSI hardware-based) or memexpd (server memory/diskstore) interchangeably to store/load lock/LVP info.  I can't see that memexpd does anything more than store/load sorts of functions, though.  It seems to have no idea of lock states, etc.  ... it just serves (load/store) raw information.

It looks to me like all the lock management, and cluster management, is done by the memexp lock modules, on a distributed basis, and not through a central "lock server".  It looks like each memexp module can communicate with its peers directly, rather than via a "lock server".  

This was a bit of a surprise to see this ... I would like to get a sanity check if anyone would be willing to take a look.

-- Ben --

Opinions are mine, not Intel's

<<winmail.dat>>