Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Le 26/10/2010 11:13, Mark Watts a écrit : > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On 10/25/2010 08:37 PM, Sam Przyswa wrote: > >> Hi, >> >> Before to swap my Zimbra and Postfix server on DRBD I have some questions: >> >> My actual setup run Zimbra (collaborative app) and Postfix in >> virtualization on 2 vserver guest and LVM, each virtual machine / is a >> logical volume. I plan to run the LVM over DRBD. >> >> 1) - First of all is the LVM on DRBD is really fiable in production mode ? >> > Using an LVM volume for DRBD backing store, and using a DRBD block > device as an LVM physical volume are both supported. > > >> 2) - In case of DRBD crash on primary is it tools to recovery the >> filesystem and what about the secondary ? >> > Assuming DRBD is in Primary/Secondary UpToDate/UpToDate, then an fsck on > the primary node is all you need. > Remember, DRBD is replicating changes to blocks. If that happens to > correspond to writes by a filesystem, then great. > > >> 3) - To migrate to the new setup I have to configure the secondary >> machine with only one machine as primary to create the logical volume >> and install the vservers. Is it a problem to start DRBD with only one >> machine before install DRBD and LVM on second machine and sync after ? >> > Single-node DRBD is fine. When you do eventually connect the second > node, a full sync will occur so you may wish to do this during scheduled > down-time or overnight. > > It's best to have drbd.conf configured for both nodes though, just don't > start DRDB on the second node until you are ready. > Thanks a lot for your reply, but what you mean by "Using an LVM volume for DRBD backing store" I'm very interested ! Sam.