Hi Paul,
Thanks for your reply.
From what you say, it's ok for me to disable the snapshot agent. That seems
the best solution for us, as we only have one named subscription, and I am
concerned about the data transfer loads of having to apply a snapshot every
week.
To answer your question of why am I reintializing, I am not! The is just
something SQL Server did, set the subscription as neefing reinitialization. I
thought it must have done this because there was an error applying the
snapshot. The subscription expiration period is set to 'never expire', as
even if the subscription is offline for a week or more we don't want it to
expire.
I do have the verbose logging for the merge agent set to 2 (that's where I
got the log from that I posted a snippet of). Unfortunately it didn't give a
reason fo rthe snapshot failure, just reported that it had failed. I suppose
the log does say 'Unable to open BCP host data-file', this could be the
reason, although I don't know what caused it to not be able to find the data
folder.
So, I think the solution to the problem I had is to disable the snapshot
agent.
Thanks for your help,
Kath
[quoted text, click to view] "Paul Ibison" wrote:
> Kath,
> in transactional and merge replication, after the initialization I disable
> the snapshot agents. In those cases where you have anonymous subscribers
> coming online and you don't control the schedule, I can understand the need
> for up to date snapshots, but most of us work with named subscribers and
> running the snapshot agents is usually redundant after the initialization.
> You can always run a manual snapshot if reinitialization is required.
> My next question is why are you reinitializing? Is this being done manually
> or is the subscription expiration period being exceeded. What is this value?
> Perhaps it needs to be increased?
> Normally when reinitializing merge, the option of uploading changes made on
> the subscriber is presented, which will avoid the issue you're seeing with
> loss of data.
> Full logging of the merge agent should present more data about why the
> process was truncated (
http://support.microsoft.com/?id=312292).
> Cheers,
> Paul Ibison SQL Server MVP,
www.replicationanswers.com >
>