Two up to date arch systems. Kernel 4.1.2 (Arch -2). 2 Mellanox MT25418 [ConnectX VPI PCIe 2.0 2.5GT/s - IB DDR / 10GigE] (rev a0) running mlx4_core driver v2.2-1 (Feb, 2014.) Both on most recent firmware for PSID MT_04A0110002, FW Version 2.9.1000. Systems directly connected, no switches. InfiniBand otherwise works great, through VERY extensive testing. Running OpenFabrics most recent releases of everything (release versions, not git versions.) Open-iscsi 2.0_873-6. targetcli-fb 2.1.fb41-1, python-rtslib-fb 2.1.fb57-1, and python-configshell-fb 1.1.fb18-1. I can't discover more than 1 iSER device working at a time. Using IPoIB lets me discover as many as I want. At the very end is a workaround - not a fix. I start with 3 disks working through iSCSI over IPoIB, with targetcli's (-fb version) ls looking like: o- / ..................................................................... [...] o- backstores .......................................................... [...] | o- block .............................................. [Storage Objects: 3] | | o- sda4 ........................ [/dev/sda4 (4.4TiB) write-thru activated] | | o- sdb4 ........................ [/dev/sdb4 (4.4TiB) write-thru activated] | | o- sdc4 ........................ [/dev/sdc4 (4.4TiB) write-thru activated] | o- fileio ............................................. [Storage Objects: 0] | o- pscsi .............................................. [Storage Objects: 0] | o- ramdisk ............................................ [Storage Objects: 0] | o- user ............................................... [Storage Objects: 0] o- iscsi ........................................................ [Targets: 3] | o- iqn.2003-01.org.linux-iscsi.terra.x8664:sn.2549ae938766 ....... [TPGs: 1] | | o- tpg1 ........................................... [no-gen-acls, no-auth] | | o- acls ...................................................... [ACLs: 1] | | | o- iqn.2005-03.org.open-iscsi:c04e8f17af18 .......... [Mapped LUNs: 1] | | | o- mapped_lun0 .............................. [lun0 block/sda4 (rw)] | | o- luns ...................................................... [LUNs: 1] | | | o- lun0 ..................................... [block/sda4 (/dev/sda4)] | | o- portals ................................................ [Portals: 1] | | o- 0.0.0.0:3260 ................................................. [OK] | o- iqn.2003-01.org.linux-iscsi.terra.x8664:sn.8518b92b052d ....... [TPGs: 1] | | o- tpg1 ........................................... [no-gen-acls, no-auth] | | o- acls ...................................................... [ACLs: 1] | | | o- iqn.2005-03.org.open-iscsi:c04e8f17af18 .......... [Mapped LUNs: 1] | | | o- mapped_lun0 .............................. [lun0 block/sdb4 (rw)] | | o- luns ...................................................... [LUNs: 1] | | | o- lun0 ..................................... [block/sdb4 (/dev/sdb4)] | | o- portals ................................................ [Portals: 1] | | o- 0.0.0.0:3260 ................................................. [OK] | o- iqn.2003-01.org.linux-iscsi.terra.x8664:sn.d4603198ba50 ....... [TPGs: 1] | o- tpg1 ........................................... [no-gen-acls, no-auth] | o- acls ...................................................... [ACLs: 1] | | o- iqn.2005-03.org.open-iscsi:c04e8f17af18 .......... [Mapped LUNs: 1] | | o- mapped_lun0 .............................. [lun0 block/sdc4 (rw)] | o- luns ...................................................... [LUNs: 1] | | o- lun0 ..................................... [block/sdc4 (/dev/sdc4)] | o- portals ................................................ [Portals: 1] | o- 0.0.0.0:3260 ................................................. [OK] o- loopback ..................................................... [Targets: 0] o- sbp .......................................................... [Targets: 0] o- srpt ......................................................... [Targets: 0] o- vhost ........................................................ [Targets: 0] On the initiator system, I clear everything. Log out via iscsiadm -m node -U all. Disconnect via iscsiadm -m discovery -t sendtargets -p IP -o delete. On the target system, i go into each of the iscsi/iqn/tpg1/portals/0.0.0.0:3260 directories and run "enable_iser true". Each time it says "iSER enable now: True". / saveconfig and exit. target-cli now changes to: | | o- 0.0.0.0:3260 ............................................... [iser] ... | | o- 0.0.0.0:3260 ............................................... [iser] ... | o- 0.0.0.0:3260 ............................................... [iser] On the initiator system, I discover via iscsiadm -m discovery -t sendtargets -p IP -I iser, and it says: iscsiadm: recv's end state machine bug? iscsiadm: Could not perform SendTargets discovery: iSCSI PDU timed out The target's dmesg added: [80296.332049] isert: isert_handle_wc: wr id ffff8800a78f1c18 status 9 vend_err 8a The initiator's dmesg added: [10868.076407] scsi host25: iSCSI Initiator over iSER [10868.078969] iser: iser_handle_wc: wr id ffff8807f7ee4000 status 1 vend_err d7 [10868.078982] connection7:0: detected conn error (1011) Now, on the target machine, if I run "enable_iser false" on two of the iqn's portals, saveconfig, and exit... Then run iscsiadm -m discovery -t sendtargets -p IP -I iser, it gives: 192.168.2.1:3260,1 iqn.2003-01.org.linux-iscsi.terra.x8664:sn.2549ae938766 Target's dmesg has nothing new, initiator's has: [11067.116617] scsi host27: iSCSI Initiator over iSER On the initiator, I can log into the node, mount it, and use it just fine. I can even discover and log into the other two nodes, using iSCSI over IPoIB rather than iSER for those 2, and use all 3. But, I can't get more than 1 iSER to discover at a time. ... Not sure if this is a kernel issue, a Mellanox issue, an OpenFabrics issue, an open-iscsi issue, or a targetcli (-fb version) issue. I found the only difference in open-iscsi's node configuration files is iface.iscsi_ifacename = iser, and iface.transport_name = iser. (Rather than default and tcp.) And, the files are called iser rather than default. If I discover the targets with the targets having enable_iser false, then stop the initiator's open-iscsi.service, update all the node config files to iser, rename them to iser, change all the targets to enable_iser true, and start the initiator's open-iscsi.service, it works. (I can log in at that point, mount them, whatever.) So, the issue is in discovery, not logging in or using. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html