Thanks very much for your quick response Jim!! But i have not understood very well.. :)
2008/2/14, Jim Dunham <[EMAIL PROTECTED]>: > Leal, > > > These are the steps that i'm following to switch back a resource > > managed by the MRSL.NONsharedDevice agent: > > > > Scenario: Did occur a failover (the system is in logging mode), the > > resource is running on the secondary node. > > > > Steps (commands): > > 1 - Bring the resourcegroup offline (clresourcegroup offline pool-rg) > > 2 - Do a reverse/update sync (sndradm -C local -g POOL -n -r -u) > > 3 - Here we have two options using the NONsharedDevice agent: > > a) Put the AVS software in logging mode again (sndradm -C > > local -g POOL -n -l) or > > b) Set the Last_Started extension to the "primary" node > > hostname (clresource set -p Last_started=primarynodename pool-rs) > > 4 - Bring the resourcegroup online on the primary node > > (clresourcegroup online -n primarynodename -M pool-rg) > > > > The question is: > > While the systems are resync the services are unavailable, the good > > point is that is a switch back scenario, so we can schedule it. But > > would be nice if i could make a reverse/sync while still using the > > resourcegroup. I mean, if there is a lot of data to resync, can be a > > long, long maintenance time. > > Once the "reverse/update sync" has been invoked, SNDR will start the > background process of resynchronizing the secondary volume to the > primary volume. Here the "direction" is: Primary <---------------- Secondary Once the "Primary" discs are being updated/resynced, this is a block order operation, and the filesystem/discs on the Primary node can NOT be accessed. Right? > While this activity is underway, access to the SNDR > primary volume can be resumed, and thus services can be made > available through a replication feature called on-demand pull. Ok, that's what i did not understand... you talk about "access" the "primary" node? How that's possible with the discs being resynchronided? I mean, i think the only consistent filesystem is on the secondary node. To mount the filesystem on the primary node i need the AVS go to Replication mode. I miss something? > > What this means is that if a block is accessed on the SNDR primary, > but it has yet to be resychronized, an on-demand request will be made > to fetch the block. Be forewarned, that when running in this mode, > performance will be less, maybe much less than running SNDR in > synchronous replication mode, since the background resynchronizing and > the on-demand pull will be competing for the same resources. About the performance, is not big deal.. today my solution to switch back has downtime. Something (slow) is betther than nothing. :) > > This mode of operation is based on the assumption that only a small > percentage of a volumes recently changed data is needed to continue > needed I/O services. > If i understand, the discs in the primary volume can be mounted "while" reverse syncing? And if for each block that needed to be accessed will be made a query to the secondary to know if it need to be get from the secondary node? is that right? Are you telling me that the switch back scenario will not have downtime? Can you explain me "editing" my "steps (commands)" above. > Jim > > Thanks again! Leal This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
