iSCSI failing with Ubuntu initiator post 9.3 Upgrade

[mloebl]

Problem: i/o errors on Ubuntu initiator with existing working config. Reverting to 9.2 resolves problem.

Description:
Upgraded from 9.2.1 to 9.3 and ran into a problem with an existing iSCSI target (this is after the latest update that came in for 9.3 with the iSCSI fix as well.)

Setup:
Private network so no discovery or targets auth (both set to None, also tried CHAP)
1 portal listening on default 3260
2 targets, each with own initiator group, no auth, and unique volume pointing to two different volumes off my RAID controller (2 distinct RAID groups)

I use the initiators auth network right now to filter who can see what (set to specific IPs), not ideal, but has been working fine for awhile.

There are 2 iSCSI initiators, one on Windows, other an Ubuntu 14.10 box. Initially I had issues where Windows box wouldn't work with None for Discovery post upgrade, so had to set up CHAP (whole separate issue), however tried this both on and off with Ubuntu with the same net result.

Failure:
Ubuntu would see the target, but was unable to mount without complaining that there were problem with the file system. Output from log:

Code:

Dec 20 14:46:24 mike kernel: [   14.942358] scsi9 : iSCSI Initiator over TCP/IP
Dec 20 14:46:24 mike kernel: [   14.944039] scsi10 : iSCSI Initiator over TCP/IP
Dec 20 14:46:24 mike kernel: [   14.945352]  connection1:0: detected conn error (1020)
Dec 20 14:46:24 mike kernel: [   15.197491] scsi 10:0:0:0: Direct-Access     FreeBSD  iSCSI Disk       0123 PQ: 0 ANSI: 6
Dec 20 14:46:24 mike kernel: [   15.198200] sd 10:0:0:0: Attached scsi generic sg4 type 0
Dec 20 14:46:24 mike kernel: [   15.198851] sd 10:0:0:0: [sdc] 3905945600 512-byte logical blocks: (1.99 TB/1.81 TiB)
Dec 20 14:46:24 mike kernel: [   15.199841] sd 10:0:0:0: [sdc] Write Protect is off
Dec 20 14:46:24 mike kernel: [   15.199843] sd 10:0:0:0: [sdc] Mode Sense: 73 00 10 08
Dec 20 14:46:24 mike kernel: [   15.200260] sd 10:0:0:0: [sdc] Write cache: enabled, read cache: enabled, supports DPO and FUA
Dec 20 14:46:24 mike kernel: [   15.227541]  sdc: unknown partition table
Dec 20 14:46:24 mike kernel: [   15.230077] sd 10:0:0:0: [sdc] Attached SCSI disk

If I ran fsck or gparted, I'd get a bunch of:

Code:

Error fsyncing/closing /dev/sdc1: Remote I/O error

or

Code:

Dec 20 14:46:24 mike kernel: [   15.644435] end_request: critical target error, dev sdc, sector 0

I was only able to get the drive mounted read only to get the data off, so for the heck of it reformatted it with the same problem. Strangely, the Windows box was even able to mount and reformat it without a problem (when I allowed it as an auth ip to connect) and did not have any of the i/o issues so definitely related to 9.3 iSCSI + Ubuntu 14.10. The back end is fine, storage array functioning properly with no errors at all as well.

Right now back running on 9.2, but can easily swap the USB key over the attempt 9.3 again.

Thanks

-Mike

 

[dlavigne]

Please create a bug report at bugs.freenas.org and post the issue number here.

 

[mloebl]

Please create a bug report at bugs.freenas.org and post the issue number here.

Thanks for the quick response;
https://bugs.freenas.org/issues/7236