Lustre doesn''t perform the failover action automatically; it is up to
you to provide a mechanism which performs the failover, say using some
cluster management tool. It will have to be responsible for starting the
OST services on the alternate node by running the lconf command there.
All Lustre does is provide tools which support implementing failover. If
you do want to set up a failover environment I would recommend setting
up an LDAP server (see the Lustre Wiki) to hold the configuration, which
can be loaded using the load_ldap.sh tool, and then you will be able to
use the lactive command to control the active OST service to server node
mappings. You will also need to implement one or more upcall scripts to
handle the failover recovery events on the MDS and client nodes, which
will call the lconf command with the --recover flag and the appropriate
arguments.
Fergal.
--
Fergal.McCarthy@HP.com <mailto:Fergal.McCarthy@HP.com>
(The contents of this message and any attachments to it are confidential
and may be legally privileged. If you have received this message in
error you should delete it from your system immediately and advise the
sender. To any recipient of this message within HP, unless otherwise
stated, you should consider this message and attachments as "HP
CONFIDENTIAL".)
-----Original Message-----
From: lustre-discuss-admin@lists.clusterfs.com
[mailto:lustre-discuss-admin@lists.clusterfs.com] On Behalf Of jfire
Sent: 04 May 2004 08:25
To: lustre-discuss@lists.clusterfs.com
Subject: [Lustre-discuss] lustre failover testing...
dear lustre.org
first thx, for your good cluster solutions.. it''s may be very useful...
:)
i have a question about ost''s failover..
i had seen lustre failover document on wiki.
it said
lmc --add ost --ost ost1 --failover --node nodeA --lov lov1 --dev
/dev/sda1
lmc --add ost --ost ost1 --failover --node nodeB --lov lov1 --dev
/dev/sdb1
so i make config file and running lustre like below..
# mds
lconf --reformat --node node1 config.xml
# ost1
lconf --reformat --node node2 config.xml
# ost2
lconf --reformat --node node3 config.xml
# client
lconf --node node4 config.xml
i allocated each ost''s 1G byte on ost device. (ost2 didn''t
create ost
device.. is it right? ost1 is fine)
anyway.. i''m copying some files on lustre mount point and going to
shutdown ost1.
but failover is not operated --; what''s problem..?
i think ost1''s ost device is shared to ost2(thru nfs or others). right?
and mds, client how to know ost is change ost1 to ost2.. i think it''s
not automatically..
waiting for reply.. thx