Hi there, does anybody have some experience with building a RAID over network? I'd like to build the following: I have five systems running Debian GNU/Linux - each with a 50GB data partition. Now I'd like to build a RAID5 or RAID6 out of those five partitions (the systems are attached at a gigabit ethernet) and mount them on the servers (using ocfs2). I tried to export the partitions with nbd and built the RAID with mdadm but the mounted RAID doesn't seem to be consistent (a file on the ocfs2 partition on node1 does not exsist on the ocfs2 partition on node2). Is this a problem with nbd, mdadm or ocfs2? Thanks in advance Carsten Hensiek -------------- next part -------------- An HTML attachment was scrubbed... URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20060911/7cc3bddd/attachment.html
As far as ocfs2 is concerned, you can use debugfs.ocfs2 to check whether you are seeing the same "physical device" on the two nodes. # echo "stats" | debugfs.ocfs2 -n /dev/sdX Compare the UUIDs. # echo "ls -l /path" | debugfs.ocfs2 -n /dev/sdX List some dir. Carsten Hensiek wrote:> > Hi there, > > does anybody have some experience with building a RAID over network? > I'd like to build the following: > > I have five systems running Debian GNU/Linux - each with a 50GB data > partition. Now I'd like to build a RAID5 or RAID6 out of those five > partitions (the systems are attached at a gigabit ethernet) and mount > them on the servers (using ocfs2). > > I tried to export the partitions with nbd and built the RAID with > mdadm but the mounted RAID doesn't seem to be consistent (a file on > the ocfs2 partition on node1 does not exsist on the ocfs2 partition on > node2). Is this a problem with nbd, mdadm or ocfs2? > > Thanks in advance > Carsten Hensiek > > ------------------------------------------------------------------------ > > _______________________________________________ > Ocfs2-users mailing list > Ocfs2-users@oss.oracle.com > http://oss.oracle.com/mailman/listinfo/ocfs2-users >
Sunil Mushran wrote:> As far as ocfs2 is concerned, you can use debugfs.ocfs2 to > check whether you are seeing the same "physical device" on > the two nodes. > > # echo "stats" | debugfs.ocfs2 -n /dev/sdX > Compare the UUIDs. > > # echo "ls -l /path" | debugfs.ocfs2 -n /dev/sdX > List some dir.Hi, thanks for you help Sunil. The output of your commands is exactly the same on all nodes. Nex Problem: To simulate crash of node2 I pulled node2?s ethernet cable and set the exported device as faulty (mdadm -f /dev/md0 /dev/nbd2). After the O2CB_HEARTBEAT_THRESHOLD time all nodes hung and crashed with a heartbeat write timeout and "link to 1 down". Thanks Carsten