This is a discussion on replication server proper shutdown within the Sybase forums, part of the Database Server Software category; --> hi with reference to an earlier post in July (see below) don't we have to quiesce the repserver, then ...
| |||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| hi with reference to an earlier post in July (see below) don't we have to quiesce the repserver, then do the suspend stuffs, like suspend log transfers, suspend connections etc, before we shutdown the server?? or does the normal shutdown do all these for us?? thanks a lot.... > Hi All, > > Will any one you will please let me know, what are the steps to shutdown > the repserver. Just a normal shutdown to the repserver will do, or I have > to kill the connections or do the quiesec. I'm using sybase 11.9.2. A regular shutdown of the rep server should be enough. |
| ||||
| s99999999s2003@yahoo.com (mike) wrote: > with reference to an earlier post in July (see below) > don't we have to quiesce the repserver, then do the suspend stuffs, like > suspend log transfers, suspend connections etc, before we shutdown the server?? > or does the normal shutdown do all these for us?? > thanks a lot.... > > > Hi All, > > > > Will any one you will please let me know, what are the steps to shutdown > > the repserver. Just a normal shutdown to the repserver will do, or I have > > to kill the connections or do the quiesec. I'm using sybase 11.9.2. > > A regular shutdown of the rep server should be enough. Although we use a shutdown protocol which first stops the RSM, then the Rep Agents, then suspends some connections, then shuts down the RepServer and, finally, the ASE holding the RSSD, I agree with other posters that all these is not mandatory. We never had a problem after an abrupt shutdown. I think that the components of a Sybase replication system are wisely coded to provide seamless reconnection after the shutdown of any component. If pursuing a clean shutdown procedure, let me explain why we proceed as we do: - We shut down the RSM first, to prevent it sending alerts about the components going down. - Then, we stop Rep Agents, following the common advice of closing the origin first and the target later. By stopping Rep Agents we prevent some messages from appearing on ASE's errorlog, warning about re-connection attemps to a stopped RepServer. On the other hand, we force ourselves to start them again later, because a RepAgent which has been explicitely stopped won't reconnect automatically. If there are chances of forgetting to start them again, better don't stop Rep Agents. - We suspend connections to some ASE servers which could be down by the time we bring RepServer up again. Personally, I think it's useless to suspend connections to servers which shall remain up. - Then, we shut the RepServer down - Finally, the ASE server which holds the RSSD. It's very important to keep it running while RepServer is up, as RepServer is almost unable to operate while the RSSD is not available. Our start-up protocol is the reverse of the shut-down. Regards, Mariano Corral |