Try inserting a delay into the iscsi init script in the beginning of the start clause. Depending on the delays in your netwrok and the NIC drivers being used I've found values of between 30 and 60 seconds worked for me.
From: linux-cluster-bounces redhat com [mailto:linux-cluster-bounces redhat com] On Behalf Of Randy Brown
Sent: Tuesday, October 23, 2007 11:40 AM
To: linux clustering
Subject: Re: [Linux-cluster] problems with iscsi storage and boot order
node.session.initial_login_retry_max = 8
I even made the value as high as 20 with no change during boot up. Then I added NETWORKDELAY=20 to /etc/sysconfig/network and I still see:
iscsiadm: Could not login session (err 4)
iscsiadm: initiator reported error (4 - encountered connection failure)
when I boot the machine. After the machine as booted, I restart iscsi, clvmd, and gfs and everything works. Sometimes I hate computers.
Bryn M. Reeves wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 gordan bobich net wrote:On Mon, 22 Oct 2007, Randy Brown wrote:Thanks, I'll try upping the retries. I am assuming this is the same thing as increasing the time value here:No. Timeouts and retries are separate settings. The problem is usually that the iSCSI subsystems tries to access the SAN before the network has fully come up.You can also tweak the network delay setting in /etc/sysconfig/network. Set the variable NETWORKDELAY to a value (in seconds) and the network scripts will sleep for this period of time after bringing all interfaces up. This should allow time for your network to be fully up before the iscsi service starts. Regards, Bryn. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iD8DBQFHHatA6YSQoMYUY94RAiGAAJ41fsuuLLvypgti/C+Ik07RCBUkrgCfUL9W tYdoAqbjAdoRr02KJKssdJQ= =nciQ -----END PGP SIGNATURE----- -- Linux-cluster mailing list Linux-cluster redhat com https://www.redhat.com/mailman/listinfo/linux-cluster