I have a T2000 running Solaris 10 11/06 with relatively current patches. It uses iSCSI with two independant TCP/IP paths to a Netapp file server. I believe I've set up multipathing correctly. It's certainly using both paths in a round-robin configuration. When I run `mpathadm show lu' it shows to paths to the disks, with both in an `OK' state.
I tested this today by disconnecting one of the ethernet cables. It seems to have switched to a single path. Here are selected messages:
Jan 3 10:02:05 pollux e1000g: [ID 801725 kern.info] NOTICE: pciex8086,105e - e1000g[2] : Adapter copper link is down. Jan 3 10:02:56 pollux scsi_vhci: [ID 734749 kern.warning] WARNING: vhci_scsi_reset 0x1 Jan 3 10:03:20 pollux iscsi: [ID 286457 kern.notice] NOTICE: iscsi connection(7) unable to connect to target iqn.1992-08.com.netapp:sn.33610487 (errno:145) Jan 3 10:06:12 pollux last message repeated 6 times Jan 3 10:06:12 pollux iscsi: [ID 328943 kern.notice] NOTICE: iscsi session(6) iqn.1992-08.com.netapp:sn.33610487 offline Jan 3 10:06:12 pollux genunix: [ID 834635 kern.info] /scsi_vhci/ssd@g60a98000433469764e4a41357149432f (ssd1) multipath status: degraded, path /iscsi (iscsi0) to target address: 0000iqn.1992-08.com.netapp:sn.336104870005,1 is offline Load balancing: round-robin
After this, `mpathadm show lu' showed one path as `OK' and the other as `unavailable'. This seems okay. However, when I reconnect the ethernet cable, I get only this:
Jan 3 10:20:23 pollux e1000g: [ID 801725 kern.info] NOTICE: pciex8086,105e - e1000g[2] : Adapter 1000Mbps full duplex copper link is up. Jan 3 10:20:24 pollux iscsi: [ID 240218 kern.notice] NOTICE: iscsi session(6) iqn.1992-08.com.netapp:sn.33610487 online
`mpathadm show lu' still shows one path unavailable. Isn't it supposed to come back automatically? How do I get scsi_vhci to recognize that it's back? Is my configuration perhaps wrong?
-- -Gary Mills- -Unix Support- -U of M Academic Computing and Networking-
Andre.Boegelsack 7 January 2008 14:17:04 [ permanent link ]
Hi Garry,
I'm not very familiar with iSCSI but normally in a FC environment the unavailable pathes come back online automatically. Perhaps you can try to offline/online the pathes again using cfgadm? I assume drvfsadm dos not work in this case?
