Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision | ||
|
sanboot:ubuntu_iscsi [2010/01/20 03:57] gruffi |
sanboot:ubuntu_iscsi [2010/08/16 20:55] (current) cinquero |
||
|---|---|---|---|
| Line 104: | Line 104: | ||
| initrd /initrd.img-2.6.28-11-generic | initrd /initrd.img-2.6.28-11-generic | ||
| quiet | quiet | ||
| + | |||
| + | WARNING: Make a backup of grub.cfg! When you upgrade your system the ubuntu installer will overwrite it. If that is the case you have to add your iscsi options again! After upgrading always check if your grub.cfg is still intact. | ||
| ===== Transfer your image to the iSCSI Target server ===== | ===== Transfer your image to the iSCSI Target server ===== | ||
| Line 197: | Line 199: | ||
| ===== FIXME: Notes from contributors ===== | ===== FIXME: Notes from contributors ===== | ||
| + | ==== Ubuntu 10.04 LTS / Lucid ==== | ||
| + | |||
| + | Disclaimer: use everything in this section at your own risk! Think for yourself. | ||
| + | |||
| + | With Lucid, it is basically the same as described above, except for: | ||
| + | |||
| + | * copying early_udev and bugfixing scripts is not necessary | ||
| + | * make sure the OS image contains its target definition under /etc/iscsi/nodes/... -- otherwise iscsid start will kill the connection upon boot. You get that target definition by mounting the iscsi target. iscsiadm seems to be creating it. | ||
| + | * use sysv-rc-conf to remove "networking" and "open-iscsi" from *all* run levels. | ||
| + | * make sure that *all* open-iscsi sysv links are gone. My experience shows that using sysv-rc-conf is *not* enough: | ||
| + | <code> | ||
| + | find /etc -name 'K*open-iscsi' | ||
| + | find /etc -name 'S*open-iscsi' | ||
| + | </code> | ||
| + | * One possibly has to repeat the previous step when upgrading/reinstalling open-iscsi. | ||
| + | * create /etc/init/open-iscsi.conf: | ||
| + | <code> | ||
| + | description "Open-iSCSI" | ||
| + | |||
| + | start on filesystem | ||
| + | # no stopping! | ||
| + | exec /etc/init.d/open-iscsi start | ||
| + | </code> | ||
| + | * Lucid detects on its own if rootfs is an iscsi target. It will prevent network shutdown on shutdown and read-only remounting of the rootfs should make sure that data gets flushed to disk. | ||
| + | * For your own convenience, edit the following line in /etc/default/grub as shown below: | ||
| + | <code> | ||
| + | GRUB_CMDLINE_LINUX_DEFAULT="ip=dhcp ISCSI_INITIATOR=<yourInitiatorName> ISCSI_TARGET_NAME=<yourTargetName> ISCSI_TARGET_IP=<yourTargetIp> ISCSI_TARGET_PORT=3260 quiet splash" | ||
| + | </code> | ||
| + | |||
| + | === Stabilization/Paranoia/Warnings === | ||
| + | |||
| + | in order to prevent data loss (for example, a corrupted dpkg db), one may do the following stuff: | ||
| + | |||
| + | * Enable data journaling. That will cut your maximum write performance in half. | ||
| + | * One may increase node.session.timeo.replacement_timeout in iscsid.conf to avoid escalation of temporary connection problems to upper layers. In *my* situation it (almost) *never* makes sense to tell the ext4 driver about I/O problems because there is no fallback. It would just give me an unclean filesystem (not unmounted properly, like a hard reset). However(!!), an unclean filesystem may be better than a totally screwed one: imagine your iSCSI target host crashes, looses some data that still has not been written to disk, and the client continues to use that target without doing a journal replay first... personally, I try to avoid that situation by never starting the iSCSI target automatically (removed from init scripts, starting it manually after killing the client machines...). IMHO the iSCSI protocol is a bit dumb. The target should detect a crash and refuse session continuation after restart... again, I'm not liable for *any* sort of data loss if you act according to these explanations. You *will* screw your data. Sooner or later. | ||
| + | * Additional warning: SAN boot using gpxe seems to not allow using a MaxSessions=1/MaxConnections=1 setting for the target, which would prevent concurrent accesses -- a very basic need for data integrity purposes. | ||
