Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-04-23 18:02:19 +0100
\ Nuno Tavares:
> Lars,
>
> 1. Nothing arrives at /tmp/drbd_1/
what makes you think so?
is it possible that you have wrong expectations?
just in case: if you mount the secondary ro, which is still possible
with 2.4 kernels, and write on the Primary, do not expect anything to
show up on your secondary, but be happy that it does not crash.
* mountiong on Secondary WILL CRASH YOUR BOX *
sooner or later.
> 3. # drbdsetup /dev/nb0 replicate
> "replicate is not a command"
> What? :)
exactly.
in 0.7 serveral things are different.
documentation is not yet up-to-date.
# ./drbdsetup
|
|USAGE: drbdsetup device command arguments options
|
|Device is usually /dev/nbX or /dev/drbd/X.
|Commands, arguments and options are:
| primary [{--human|-h}] [{--do-what-I-say|-d}]
| [{--timeout-expired|-t}]
| secondary
| wait_sync [{--time|-t} val]
| wait_connect [{--wfc-timeout|-t} val] [{--degr-wfc-timeout|-d} val]
| invalidate
| invalidate_remote
| syncer [{--use-csums|-c}] [{--skip-sync|-k}] [{--group|-g} val]
| [{--rate|-r} val] [{--al-extents|-e} val]
| down
| detach
| net local_addr remote_addr protocol [{--timeout|-t} val]
| [{--max-epoch-size|-e} val] [{--max-buffers|-b} val]
| [{--connect-int|-c} val] [{--ping-int|-i} val]
| [{--sndbuf-size|-S} val]
| disk lower_dev meta_data_dev meta_data_index [{--size|-d} val]
| [{--on-io-error|-e} val]
| resize [{--size|-d} val]
| disconnect
| show
|
|Available on-io-error handlers: pass_on, panic, detach
|
|Version: 0.7-pre6 (api:72)
> I really liked what you are doing here, but I need a solution *fast*.
> Still can't get drbd 0.6.12/0.7 to work, but had not much time to
> experiment. I'd like to do it this evening. The problem is that I
> setup both servers and cat /proc/cpu keeps showing 1: has unConfigured.
> iptables is off. I had to do something wrong, will look at it tonight.
proc/cpu ? 1: has unConfigure ? what are you talking about ??
> What I need is some solution that gets this possible: I have server1
> and server2, both with a local filesystem subtree, which should be
> identical in both servers and sync'ed in real time (as fast as possible).
Nope. This is not what DRBD is supposed to do.
> Suppose this: you login to server1 and "echo SOMETHING > ~/TEST". Logout
> and immediately login to server2. When you do "cat ~/TEST" you may see 1)
> that the file exists and 2) look at its contents (SOMETHING).
use a shared filesystem, NFS, SMB, anything.
put it on a fileserver.
fileserver itself may be a HA-cluster,
build with DRBD or any other shared storage.
Lars Ellenberg