|
Hi, I just recently setup an Openfiler 1.1.1 cluster using two
IBM X345 servers and 6x 146GB RAID 5 disks in each one. I used the
Openfiler HA and Replication Howto, which by the way, isn’t anywhere on
this site, but I was somehow able to find it in a google spider cache of
sources.voluna.com. Could you please publish that document somewhere? It’s
the only setup instructions I was able to find for Openfiler and
DRBD/Heartbeat. I wanted to provide some feedback, because there are some
areas of this document that are wrong or out of date. -
In section 2.1.4, step 5, I had to
reboot the server before I was able to do a “drbadm adjust vg0_drbd”. -
In section 2.1.5, the Heartbeat
configuration asks you to put your files in /etc/ha.d directory. The
default configuration of heartbeat that comes with Openfiler 1.1 has heartbeat
look at /usr/local/etc/ha.d for configuration files. This problem is
compounded by the fact that the /etc/ha.d/resource.d/drbddisk file is NOT
located in /usr/local/etc/ha.d/resource.d. In order to fix this, I had to
copy /etc/ha.d/resource.d/drbddisk to /usr/local/etc/ha.d/resource.d/, then
delete the /etc/ha.d directory and symlink /usr/local/etc/ha.d to /etc/ha.d -
In section 2.1.8, step 6, the order
for symlinking is backwards (I think). Either it’s backwards, or
every other symlinking instruction is backwards. I think the intent is to
have all other configurations pull from /cluster_metadata, while the rsync.xml
file is actually pulled from /opt/openfiler.local/etc on each node (unique file
on each node). Is this correct? -
I also had to manually create an
/etc/ha.d/haresources file. It got overwritten by openfiler at a later
date, but in order to start heartbeat properly, I had to do this. Things I’m confused about: -
Openfiler wants to build its own
haresources file, which is fine, but when I overrode my default Netbios name in
Samba config (gave it a name that is assigned to the floating IP address of the
cluster) it screwed up my haresources file by changing that to the same field
as my Netbios name. Is there a way to prevent this? I want the
Netbios name to be generic (not tied to either node name), so that it can
failover properly. -
How do I manually failover from one
node to the other in order to do maintenance? How do I failback? I’m new to heartbeat and drbd, but so far it seems to
be working ok. Thanks in advance for your help. Regards, Luke Youngblood Linux Administrator Cardean Learning Group (203) 973-2213 http://www.cardeanlearninggroup.com |
_______________________________________________ Openfiler-users mailing list [email protected] https://lists.openfiler.com/mailman/listinfo/openfiler-users
