Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi, > Hi, > > DRBD being a kernel module, it does not have an actual "service" to > restart. sure. > "drbdadm adjust all" should be enough in most cases. > If you changed something to the network config, it's safer to "drbdadm > disconnect" the resources first, adjust the nodes and then reconnect. Perfect, thanks, that was what I was looking for in a totally wrong corner. Best, Timo > This can be done without interrupting production. > Just remember to have the exact same config on both nodes. > > It's useless to unload the module unless you upgrade it. > > Check your distro's script to know what "service drbd restart" does, > but probably it will stop production when it's not necessary. > > Lionel Sausin. > > Le 02/09/2013 08:18, lists at riscworks.net a écrit : >> Hi list, >> >> I'm about to replace a node (running as secondary) already being >> orbited by its own debris by a new machine. >> >> Having edited the config on the primary node (replaced the lines of >> the old node by the appropriate settings for the new one, i.e., IP >> address) I see that a 'service drbd reload' does not deploy the new >> config -- the (new) secondary still is not able to connect to the box >> running as primary. Networking is perfectly alright between the >> machines, I assume the primary just not being aware of the (new) >> secondary right now. >> >> As it's an in-production setup, I'm a bit hesitating to issue a >> 'service drbd restart' on the primary, additionally because it's >> highly speculative for me if it'd fix the problem. So, is there the >> need to *restart* DRBD to pick up the new config? >> >> I'm running latest CentOS 6.4 x86_64 with DRBD 8.3 from elrepo: >> >> kmod-drbd83-8.3.15-1.el6_3.elrepo.x86_64 >> drbd83-utils-8.3.15-1.el6.elrepo.x86_64 >> >> Thanks, >> >> Timo