On Tue, 2009-04-14 at 11:55 -0500, Mike Christie wrote: > Oh yeah, can you do a real quick test? Can you just do a dd or whatver > IO test you like to use to some EQL iscsi lun (forget lvm), then force > the load balancing. On the initiator we should see a logout message, > then we should see a message about the connection is operational again. > You should not see any IO errors. Then could you retry with lvm on top > of iscsi?
I should apologize for the bad copy/paste job in my initial email. I forgot to include the iscsid logout and reconnect that was indeed in the logs. So after upgrading the equallogics to firmware v4.0.6 last night I tried replicating this as you requested by doing 2 runs of dd to different volumes which after a brief period seemed to trigger the load balancing. I ran each test twice. Results: dd direct to raw device - load balanced fine: Apr 16 11:46:46 constellation iscsid: Target requests logout within 3 seconds for connection Apr 16 11:46:47 constellation iscsid: received iferror -38 Apr 16 11:46:47 constellation last message repeated 2 times Apr 16 11:46:47 constellation iscsid: connection22:0 is operational after recovery (1 attempts) dd to a file on device formatted with xfs - load balanced fine: Apr 16 12:01:47 constellation iscsid: Target requests logout within 3 seconds for connection Apr 16 12:01:48 constellation iscsid: received iferror -38 Apr 16 12:01:48 constellation last message repeated 2 times Apr 16 12:01:48 constellation iscsid: connection25:0 is operational after recovery (1 attempts) dd to a file on device formatted with xfs on top of lvm - load balanced fine: Apr 16 12:19:46 constellation kernel: [1388338.439405] iscsi_tcp: dropping R2T itt 53 in recovery... Apr 16 12:19:47 constellation iscsid: Target requests logout within 3 seconds for connection Apr 16 12:19:48 constellation iscsid: received iferror -38 Apr 16 12:19:48 constellation last message repeated 2 times Apr 16 12:19:48 constellation iscsid: connection26:0 is operational after recovery (1 attempts) The only wrinkle is the extra 'dropping R2T' message during the test with lvm. It also took far far longer to trigger the load balancing, but I doubt that's anything to do with iscsi. Oh and this is still on CentOS 5.2, haven't yet upgraded. -- Matthew Kent \ SA \ bravenet.com --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to open-iscsi@googlegroups.com To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/open-iscsi -~----------~----~----~----~------~----~------~--~---
