I've set up a Linux file server for the municipal offices of my town (as a volunteer consultant and Linux evangelist). It's working fine -- Samba makes it look like a Win machine to the rest of the world. But now I'm interested in setting up a second machine as a redundant backup to the first one. The intent is that if one goes down, the other one will take over with little or no lost data and the shift will be imperceptible to the network users. What software should I be looking at to accomplish this? Paul
On Wed, 2005-11-30 at 12:56 -0500, Paul W. Abrahams wrote:
Two packages you might want to take a look at are rsync for duplicating data between machines and heartbeat for detecting if a machine goes does down (for auto fail over I think). There is another package for syncing two machines in near real time, it treats the second as a block device for data backup, but I can't remember the name of the package (old age setting in I think but I can't remember). -- Ken Schneider UNIX since 1989, linux since 1994, SuSE since 1998
Am Mittwoch, 30. November 2005 20:39 schrieb Ken Schneider:
Do you mean drbd (www.drbd.org)? Perhaps the following presentation is of interest: http://tinyurl.com/bsxzs Cheers Oliver
Paul W. Abrahams wrote:
What software should I be looking at to accomplish this?
http://www.linuxvirtualserver.org/ Unless you're using a shared/switched storage system, you'll probably also want to check out http://www.drbd.org/ /Per Jessen, Zürich -- http://www.spamchek.com/ - managed anti-spam and anti-virus solution.
On 11/30/05, Per Jessen <per@computer.org> wrote:
The SUSE HA Cluster guru (Lars Marowsky-Bree) hangs out on Linux-HA (heartbeat). For SLES he officially supports Linux-HA and DRBD so i assume he is the best resource you will come across. Thus if I were you, I would wander over to the Linux-HA mailing list. Actual answers: Assuming you are wanting to setup a redundant Samba server, there are some issues at failover time. DRBD ensures the data is duplicated and heartbeat will move the IP for you and manage DRBD, but any open sockets get closed. As I understand it, this is also a problem for pure Windows failover clusters. Apparently a lot of windows software assumes that sockets are unreliable and simple reconnect on a dropped socket, but it is not a seamless transition. Again, I think the best place to ask is the Linux-HA mailing list. Greg -- Greg Freemyer The Norcross Group Forensics for the 21st Century
On Wednesday 30 November 2005 18:56, Paul W. Abrahams wrote:
Basically, what I setup is: 1) a cheap BIG ide disk on server mounted on dir /backup. 2) configure storebackup to backup all windows shares, keeping: - 7 daily backups - 4 end of week backups - 3 end of month backups. 3) share the backup directory via samba. This allows the users to get old versions of their files, directly from the backup share, thereby recovering from the "Uppps deleted wrong file", and "Shit! Word lost the last 10 pages of the document" errors. T'is simple and a very visible benefit to your users! Jerry
On Wed, 2005-11-30 at 12:56 -0500, Paul W. Abrahams wrote:
Two packages you might want to take a look at are rsync for duplicating data between machines and heartbeat for detecting if a machine goes does down (for auto fail over I think). There is another package for syncing two machines in near real time, it treats the second as a block device for data backup, but I can't remember the name of the package (old age setting in I think but I can't remember). -- Ken Schneider UNIX since 1989, linux since 1994, SuSE since 1998
Am Mittwoch, 30. November 2005 20:39 schrieb Ken Schneider:
Do you mean drbd (www.drbd.org)? Perhaps the following presentation is of interest: http://tinyurl.com/bsxzs Cheers Oliver
Paul W. Abrahams wrote:
What software should I be looking at to accomplish this?
http://www.linuxvirtualserver.org/ Unless you're using a shared/switched storage system, you'll probably also want to check out http://www.drbd.org/ /Per Jessen, Zürich -- http://www.spamchek.com/ - managed anti-spam and anti-virus solution.
On 11/30/05, Per Jessen <per@computer.org> wrote:
The SUSE HA Cluster guru (Lars Marowsky-Bree) hangs out on Linux-HA (heartbeat). For SLES he officially supports Linux-HA and DRBD so i assume he is the best resource you will come across. Thus if I were you, I would wander over to the Linux-HA mailing list. Actual answers: Assuming you are wanting to setup a redundant Samba server, there are some issues at failover time. DRBD ensures the data is duplicated and heartbeat will move the IP for you and manage DRBD, but any open sockets get closed. As I understand it, this is also a problem for pure Windows failover clusters. Apparently a lot of windows software assumes that sockets are unreliable and simple reconnect on a dropped socket, but it is not a seamless transition. Again, I think the best place to ask is the Linux-HA mailing list. Greg -- Greg Freemyer The Norcross Group Forensics for the 21st Century
On Wednesday 30 November 2005 18:56, Paul W. Abrahams wrote:
Basically, what I setup is: 1) a cheap BIG ide disk on server mounted on dir /backup. 2) configure storebackup to backup all windows shares, keeping: - 7 daily backups - 4 end of week backups - 3 end of month backups. 3) share the backup directory via samba. This allows the users to get old versions of their files, directly from the backup share, thereby recovering from the "Uppps deleted wrong file", and "Shit! Word lost the last 10 pages of the document" errors. T'is simple and a very visible benefit to your users! Jerry
participants (6)
-
Greg Freemyer -
Jerry Westrick -
Ken Schneider -
Oliver Wittenburg -
Paul W. Abrahams -
Per Jessen