| Author |
Topic |
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-19 : 02:59:18
|
| Hi,The topic may seem odd but this is the situation. We have an active/active cluster system. At some point, due to 'technical' and 'human' errors as our techsupport kept stating (coz they didn't know what to do or how the problem started) and we can't do anything about it but cope up, the 2nd node of the cluster failed. Rejoining it failed, in the end, it's not possible to rejoin it again.So, here's the plan1. Transfer Load from Node2 to another server, separate from the cluster (successfull). This node was used for reports so there was a window of opportunity for the transfer.2. Node1 is the primary server and the strategy for #1 step will not do.Other Info: Node1 is supporting multiple replication to different servers.Questions: 1. Will it be possible to setup logshipping on Node1 (there are more than 10 DBs)?2. How do I backup the replication information and just change connection information or do I need to recreate? |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-19 : 03:21:00
|
In addition, will i better off doing replication of the entire server instead of logshipping? We're just talking about a couple of hours of replication, just in time to redirect applications and test for stability. Then just stop replication.What of the cluster you may now ask, we'll rebuild it the old-fashion way, reconfigure |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-19 : 04:27:34
|
| Follow up questions, 1. if I do use logshipping and the cluster was rebuild successfully, can I perform reverse roles?2. if I don't intend to reverse roles, how do I perform a 'clean-up'?thanks in advance |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-25 : 03:24:07
|
| Please help...Any advice is good advice |
 |
|
|
mr_mist
Grunnio
1870 Posts |
Posted - 2004-08-25 : 04:52:01
|
| I'm still not entirely sure what the question is.You can set up log shipping on as many or as few databases as you desire.You can reverse roles, if the setup is designed for it.I'm not sure what you're attempting to achieve.Replication is not for DR.-------Moo. :) |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-25 : 05:06:21
|
| Hi Moo,I'm trying to transfer the load from a cluster node to a non-clustered server.What's the best approach?Will logshipping be the same whether the primary machine is clustered or not?thanks... |
 |
|
|
mr_mist
Grunnio
1870 Posts |
Posted - 2004-08-25 : 05:17:57
|
| Backup / Restore?And yes, it's the same.-------Moo. :) |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-25 : 05:31:20
|
Backup and Restore will be my last resort, this is a live server and estimated downtime should be an hour max.As previously posted:> The server replicates more than one db to different subscribers> It houses more than 10 dbsI don't have a spare cluster system to test the logshipping scenario that's why I posted this hoping someone had encountered the same issue.Do you think i'd be better off doing a backup and restore procedure during the downtime instead of logshipping? The database to be logshipped is also the publisher database |
 |
|
|
mr_mist
Grunnio
1870 Posts |
Posted - 2004-08-25 : 05:44:06
|
| Hmm. I do not do replication so I can't comment on how they might co-exist.The thing with log shipping is that it has to start from a complete backup, so you have no choice, you're only shipping the changes. So, unless you already have a point from which to start, you have to take a full backup.-------Moo. :) |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-08-25 : 06:11:01
|
Yes. I was thinking in the line of less downtime if I do the logshipping instead of one time full backup/restore.My worry is if I do backup/restore, how do i backup replication settings. I already once restored the database in question and replication setting was not backed up.If only I could detach the databases and just add the harddisk to the other server (that'd make the task easier) but the problem is that it's in a cluster.Anyway, thanks for the replies though, it's better brainstorming with someone other than yourself. I still have a week to plan this out. |
 |
|
|
|