Hello all,
I''m making some tests using XEN 3.1. Now I have a machine exporting
iSCSI blocks (could be considered a SAN) devices to several dom0s that
run different machines. After making some live migrations I would like
to automate the process for being able to offer High Availability. What
I want to do is find a way to heartbeat machines in a ring structure, so
that if a machine crashes the one that is "pinging" will be able to
cope
with its work temporarily till replacement. I''ve been reading for a
while about HA on linux machines, but I don''t find a solution for this.
Let me explain it easily:
I need some sort of daemon that detects if the other host is down and
let me execute a custom script to live migrate machines. Heartbeat,
Keepalive or ucarp, as far as I know, don''t do this. They first detect
when the host is down and after they activate a virtual IP address. If
the service that they make HA is stateless, let''s say DNS, you
don''t
need to do any syncing between disk, if it is a Database you have to do
DRBD.
The issue here is that the storing is distributed, so I don''t have to
sync block devices. Is there a way to configure one of these services to
be able to let me execute what I want? If there is I haven''t been able
to find it.
Does anyone know what open software solution could be used to do HA this
way?
Thank you very much in advanced for your time and effort, Regards
Miguel Araujo
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users
I''m being using a setup similar to the one described here - http://www.certifried.com/?q=node/4 - with great success for months. All the Best, António Miguel Araujo wrote:> Hello all, > > I''m making some tests using XEN 3.1. Now I have a machine exporting > iSCSI blocks (could be considered a SAN) devices to several dom0s that > run different machines. After making some live migrations I would like > to automate the process for being able to offer High Availability. What > I want to do is find a way to heartbeat machines in a ring structure, so > that if a machine crashes the one that is "pinging" will be able to cope > with its work temporarily till replacement. I''ve been reading for a > while about HA on linux machines, but I don''t find a solution for this. > Let me explain it easily: > > I need some sort of daemon that detects if the other host is down and > let me execute a custom script to live migrate machines. Heartbeat, > Keepalive or ucarp, as far as I know, don''t do this. They first detect > when the host is down and after they activate a virtual IP address. If > the service that they make HA is stateless, let''s say DNS, you don''t > need to do any syncing between disk, if it is a Database you have to do > DRBD. > > The issue here is that the storing is distributed, so I don''t have to > sync block devices. Is there a way to configure one of these services to > be able to let me execute what I want? If there is I haven''t been able > to find it. > > Does anyone know what open software solution could be used to do HA this > way? > > Thank you very much in advanced for your time and effort, Regards > > Miguel Araujo_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> I need some sort of daemon that detects if the other host is down and > let me execute a custom script to live migrate machines. Heartbeat, > Keepalive or ucarp, as far as I know, don''t do this. They first detect > when the host is down and after they activate a virtual IP address. If > the service that they make HA is stateless, let''s say DNS, you don''t > need to do any syncing between disk, if it is a Database you have to do > DRBD.Heartbeat *does* allow scripting -- in fact, its entire HA startup/shutdown is shell script-based. You can do about anything you want with heartbeat and a little creativity. DRDB is not sufficient for database mirroring. DRDB is not, in and of itself, a HA/clustering solution; all it does is disk mirroring. In the event of a dom0 failure, you can assume your domU''s have failed. STONITH the downed dom0, set up the storage on the backup dom0 machine, start up the domU''s. John -- John Madden Sr. UNIX Systems Engineer Ivy Tech Community College of Indiana jmadden@ivytech.edu _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
John Madden escribió:>> I need some sort of daemon that detects if the other host is down and >> let me execute a custom script to live migrate machines. Heartbeat, >> Keepalive or ucarp, as far as I know, don''t do this. They first detect >> when the host is down and after they activate a virtual IP address. If >> the service that they make HA is stateless, let''s say DNS, you don''t >> need to do any syncing between disk, if it is a Database you have to do >> DRBD. >> > > Heartbeat *does* allow scripting -- in fact, its entire HA > startup/shutdown is shell script-based. You can do about anything you > want with heartbeat and a little creativity. > > DRDB is not sufficient for database mirroring. DRDB is not, in and of > itself, a HA/clustering solution; all it does is disk mirroring. > > In the event of a dom0 failure, you can assume your domU''s have failed. > STONITH the downed dom0, set up the storage on the backup dom0 machine, > start up the domU''s. > > John > >Thanks both for your answers, Sorry, but I did not find any about running shell scripts from Hearbeat, I will keep reading about it (I must have jump it), thanks for the advice. Also I''m going to read your link António. I know what DRBD does, what I was saying is that if I don''t have a way to synchronize data between nodes (let''s say DRBD) it would be impossible to make heartbeat with a stateful service. I know you still loose your RAM if the master node crashes. For sure, DRBD it is not an HA solution, for that you would need DRBD (or other way) + Heartbeating. Thanks a lot, Miguel _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> I know what DRBD does, what I was saying is that if I don''t have a way > to synchronize data between nodes (let''s say DRBD) it would be > impossible to make heartbeat with a stateful service. I know you still > loose your RAM if the master node crashes. For sure, DRBD it is not an > HA solution, for that you would need DRBD (or other way) + Heartbeating.Actually, you''ve got iSCSI, so you don''t need DRDB, just mount the target on the other dom0. John -- John Madden Sr. UNIX Systems Engineer Ivy Tech Community College of Indiana jmadden@ivytech.edu _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
John Madden escribió:>> I know what DRBD does, what I was saying is that if I don''t have a way >> to synchronize data between nodes (let''s say DRBD) it would be >> impossible to make heartbeat with a stateful service. I know you still >> loose your RAM if the master node crashes. For sure, DRBD it is not an >> HA solution, for that you would need DRBD (or other way) + Heartbeating. >> > > Actually, you''ve got iSCSI, so you don''t need DRDB, just mount the > target on the other dom0. > > John >Right, that''s why I want to run a shell script, to mount it on the other and start it. But the script has to check more things and change some permissions. Miguel _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users