>-----Original Message-----
>Replication is a process which is designed with a high
>degree of disconnectedness in mind. In other words to be
>fault tolerant. More than 90% of the errors you will
ever
>see can be solved by restarting the particular jobs.
>
>So your experience is typical.
>
>The handholding/monitoring can be extensive depending on
>your application. However what you can do is have the
>agents run continously, but also restart every 10 minutes
>or so. this way when they encounter a transient error,
>the next time they restart they will succeed. There are
>replication alerts that can be used for monitoring. And
it
>is possible to read the distribution history tables in
the
>distribution database to roll your own alerting functions
>and have thresholds associated with them to capture
>constantly failing jobs.
>
>If you are down for an extended period of time,
>replication will prompt you to do a resync, which will
>resend all the data from your publisher to your
>subscribers again.
>
>Normally this is a pretty safe procedure.
>
>>-----Original Message-----
>>We are getting ready to roll out our first replication
>project to a small
>>test group.
>>
>>I think we have everything set up, and it is working
>pretty well.
>>
>>I am a bit nervous though, as we occasionally observe
>errors in the
>>Replication monitor, and the entire process stops.
>Depending on the error,
>>we are usually able to fumble around and get it
>restarted, but most of the
>>errors are somewhat cryptic, and we seem to get
different
>errors depending
>>upon what broke.
>>
>>My question is that in general, is SQL replication
pretty
>stable?
>>
>>What kind of handholding/monitoring is usually required?
>>
>>Are there good resources for troubleshooting errors as
>they come up?
>>
>>And worse case, if replication were to go down for a few
>days, or the tables
>>on either end were to get
trashed/corrupted/inconsistent,
>are there
>>tools/procedures available to either automatically
resync
>the data, or will
>>we be facing tons of manual cleanup?
>>
>>We are doing some pretty simple stuff, mainly updateable
>transactional
>>replication, and don't anticipate a lot of conflicts,
but
>are wanting to get
>>a feel for how much support time we may need to allocate
>to this project as
>>it rolls out.
>>
>>Any advice or insight is appreciated....
>>
>>Thanks.
>>
>>
>>.
>>
>.