[Bug 1108623] iSCSI initiator delays boot for ~2 mins
http://bugzilla.suse.com/show_bug.cgi?id=1108623 http://bugzilla.suse.com/show_bug.cgi?id=1108623#c4 Lee Duncan <lduncan@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |CONFIRMED Flags| |needinfo?(develop7@develop7 | |.info) --- Comment #4 from Lee Duncan <lduncan@suse.com> --- I suspect there is more happening here that is not described. Have you set up any iSCSI targets? If so, have you connected to them? What is the output of "systemctl status iscsid.socket"? Same question for "iscsid.service" and "iscsid.socket". What is the output of "iscsiadm -m node"? It looks like you must have nodes that you have discovered, and you have set them up to automatically login on boot. The problem is that there are two services here: iscsid (the daemon) and iscsi (the login service, layered on top of iscsid). iscsi.service (login/logout to/from targets) ^ | iscsid.service (the iscsid daemon) <-- iscsid.socket (socket activation) The iscsid.socket unit is there to automatically start up iscsid.service if somebody (anybody) tries to communicate with the iscsid daemon. So even if iscsid.service is disabled (which it is, by default), if iscsid.socket is enabled (as it is by default), running any iscsiadm command that needs the daemon will start it up. [By the way, these default values are the same as "start on demand", so I'm not sure what Yast is doing here when you make that choice, but I suspect it does nothing.] On top of this, you have an _optional_ service to login to targets at startup that you have designated as "node.startup = automatic". By default node.startup is set to "manual". So, if you have any nodes configured, can you show their attributes, using:
iscsiadm -m node --op show
On my tumbleweed system I am not able to reproduce your symptoms so far. -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com