I have a client with multiple sites. We currently use NetWare based DA's on each major site, using a single scope (and using the 'set SLPDA NDS Forward Flag = off' on all but one DA to avoid eDir collsions!;-)

As we plan our OES2 migration we need to consider the replacement of the NetWare eDirectory integrated SLP DA service with openslp.... So far it doesn't look too good in the 'feature parity' side on things!

Concerns are about resiliency:
- WAN link failures - major sites need to continue to function when links fail
- SLP DA server failures, reboots, etc.

Firstly how do you ensure that each DA has the same list of services with openslp? Sure, you specify each DA manually on anything that registers a service, but if the slpd service is restarted how quickly does this become fully populated? How do you get it 'in sync' with the other DA(s)?*

Secondly, along same lines, if a DA has to be restarted (or even relocated due to server failure) how quickly does it take to become functional? (i.e. re-populated with services)

From my readings so far it sounds like openslp da doesn't cache info to the file system (i.e. service list lives in memory)
(What we really need in a pervasive cache... something like edirectory!! ;-)

Has anyone done anything funky to get around this issue?
*Could it be possible to NCS cluster enable the service with shared disk space and then, using the unload and load scripts, use something like slptool to extract service entries and write them out to a file on the shared disk and re-import them on service startup? (could slp.reg be used for this?!?)

Just wondering how people are getting on with slp in multi site, pure OES linux environments and how they cope with line and service failures

Any thoughts and suggestions welcome!