Hi all, after scouring the conversations here and trying a bunch of things, I've decided to stop lurking and post this before I give up on ESOS
Looking to spin up a vSphere lab using fiber channel for storage which I did using openfiler many years ago. I'm running ESOS 4.1.9 with a Qlogic QLE2464 HBA, point-to-point connections to 4 ESXi hosts. 3 of the ESXi hosts are using Emulex LPe11000 HBAs and one has a Qlogic QLE2462 HBA. SCST config attached, running vdisk_blockio without LVM or filesystem, trying to keep it as basic as possible
When I enable a target on ESOS and add the initiator, the Emulex HBAs cause the ESXi hosts to freeze up. Once I disable the target on the ESOS side, it comes back. Trying with the QLogic HBA in case there's a problem with Emulex initiators, it doesn't freeze but I never see the path or device appear on the ESXi side. ESOS sees the session with a couple commands and that's it. Nothing in the /var/log/messages on ESOS during this time that looks like an error
With all targets disabled I do see this in the messages log, assuming it's normal?
Oct 6 13:37:58 storage kernel: [73674.336906] Loop down - aborting ISP.
Oct 6 13:37:58 storage kernel: [73674.336937] qla2xxx [0000:06:00.1]-4807:8:
Oct 6 13:37:58 storage kernel: [73674.336938] ISP abort scheduled.
Oct 6 13:37:58 storage kernel: [73674.336941] qla2xxx [0000:06:00.1]-00af:8:
Oct 6 13:37:58 storage kernel: [73674.336942] Performing ISP error recovery - ha=000000004c7640e4.
Oct 6 13:37:59 storage kernel: [73674.597987] qla2x00_init_firmware(8): Neither initiator, nor target mode enabled
Oct 6 13:37:59 storage kernel: [73674.597988] qla2xxx [0000:06:00.1]-8832:8:
Oct 6 13:37:59 storage kernel: [73674.597989] Configure loop done, status = 0x0.
Oct 6 13:37:59 storage kernel: [73674.609913] qla2xxx [0000:06:00.1]-8822:8:
Oct 6 13:37:59 storage kernel: [73674.609915] qla2x00_abort_isp succeeded.
Oct 6 13:37:59 storage kernel: [73674.609917] qla2xxx [0000:06:00.1]-4808:8:
Oct 6 13:37:59 storage kernel: [73674.609918] ISP abort end.
Oct 6 13:37:59 storage kernel: [73674.609922] qla2xxx [0000:06:00.1]-4800:8:
Oct 6 13:37:59 storage kernel: [73674.609923] DPC handler sleeping.
Some things I have tried:
- LVM on the disk, adding a filesystem and a virtual disk with vdisk_fileio device type
- vdisk_fileio as block device
- Added an iSCSI target and configured on the ESXi side, this seems to work fine but I'm on 1GB ethernet and no 10GB available, thus the need for FC
Any advice would be great, really don't want to go back to openfiler given it's ancient and the whole rpath dead project issue. Always felt very hacky on the fiber channel side, performance issues, etc