Input/output error /dev/nst1 doesn't respond to a SCSI INQUIRY

Troubleshooting
1

Hi,

We recently upgraded OS Centos 7 to Rocky 8, now we seeing error on our server which attached to tape library. Its not able to query to scsi device. Everything works find on Centos but not on Rocky.

I was googling around and learned its kinda kernel module or Linux ST driver issue. anyone else seen this error. Uploaded some relevant information.

~]# sg_logs -a /dev/nst1
inquiry: pass-through os error: Input/output error
/dev/nst1 doesn’t respond to a SCSI INQUIRY
sg_logs failed: Some other error

~]# sg_map
device /dev/nst0 failed on scsi ioctl(idlun), skip: Input/output error
device /dev/nst1 failed on scsi ioctl(idlun), skip: Input/output error
device /dev/nst2 failed on scsi ioctl(idlun), skip: Input/output error
device /dev/nst3 failed on scsi ioctl(idlun), skip: Input/output error
device /dev/nst4 failed on scsi ioctl(idlun), skip: Input/output error
/dev/sg0 /dev/sda
/dev/sg1 /dev/sdb
/dev/sg2
/dev/sg3
/dev/sg4
/dev/sg5
/dev/sg6
/dev/sg7
/dev/sg8
~]# lspci -v | grep Fibre
3b:00.0 Fibre Channel: Emulex Corporation LPe12000 Series 8Gb Fibre Channel Adapter (rev 03)
3b:00.1 Fibre Channel: Emulex Corporation LPe12000 Series 8Gb Fibre Channel Adapter (rev 03)
5e:00.0 Fibre Channel: Emulex Corporation LPe12000 Series 8Gb Fibre Channel Adapter (rev 03)
5e:00.1 Fibre Channel: Emulex Corporation LPe12000 Series 8Gb Fibre Channel Adapter (rev 03)
af:00.0 Fibre Channel: QLogic Corp. ISP2532-based 8Gb Fibre Channel to PCI Express HBA (rev 02)
af:00.1 Fibre Channel: QLogic Corp. ISP2532-based 8Gb Fibre Channel to PCI Express HBA (rev 02)
~]# modinfo lpfc | head -5
filename: /lib/modules/4.18.0-553.5.1.el8_10.x86_64/kernel/drivers/scsi/lpfc/lpfc.ko.xz
version: 0:14.0.0.21
author: Broadcom
description: Emulex LightPulse Fibre Channel SCSI driver 14.0.0.21
license: GPL

~]# lsscsi -g | grep -v disk
[9:0:0:0] tape IBM ULT3580-TD6 KAJ8 /dev/st1 /dev/sg4
[9:0:1:0] tape IBM ULT3580-TD6 KAJ8 /dev/st3 /dev/sg6
[9:0:1:1] mediumx IBM 03584L32 1B00 - /dev/sg8
[15:0:0:0] tape IBM ULT3580-TD9 R3G0 /dev/st0 /dev/sg2
[15:0:0:1] mediumx IBM 03584L32 1B00 - /dev/sg3
[21:0:0:0] tape IBM ULT3580-TD6 KAJ8 /dev/st2 /dev/sg5
[21:0:1:0] tape IBM ULT3580-TD6 KAJ8 /dev/st4 /dev/sg7

~]# mt -f /dev/nst0 status
/dev/nst0: Input/output error

~]# lsmod |grep st
st 65536 0

~]# lsmod |grep lpfc
lpfc 1224704 0
nvmet_fc 40960 1 lpfc
nvme_fc 53248 2 lpfc,qla2xxx
scsi_transport_fc 86016 2 lpfc,qla2xxx

~]# uname -a
Linux ds01 4.18.0-553.5.1.el8_10.x86_64 #1 SMP Thu Jun 6 09:41:19 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux

~]# modinfo qla2xxx | grep version
version: 10.02.09.100-k
rhelversion: 8.10
srcversion: E69F6B726F4D3EE170999E5
vermagic: 4.18.0-553.5.1.el8_10.x86_64 SMP mod_unload modversions

2

Dear guest_joshi,

before we go any further we would appreciate if you could tell us who you are and with which institute you are affiliated.

Next thing is that at CERN, we do not have experience with Rocky 8.

Following point is that it looks like you have two HBAs installed on the server - the Emulex one with 4 ports and then QLogic one with 2 ports. Is that correct?

Next - your /dev/nst0 is most likely connected to LTO6 tape drive. Is that correct? If yes - please note that at CERN, we do not have experience with LTO6 tape drive.

Bottom line, you have a configuration with which we no experience. Maybe someone else from the community can help?

Best regards,

Vladimir Bahyl
CERN

3

My name is Joshi I work with nAG and provide support to HPC customers.

Following point is that it looks like you have two HBAs installed on the server - the Emulex one with 4 ports and then QLogic one with 2 ports. Is that correct?

Actually there are 3 HBAs, 2 dual port Emulex and one dual port QLogic. Those HBA connected to the tape devices.

Thanks for checking on my request.

4

Dear Joshi,

thank you for your reply and explanations that you work for nAG.

Please note that this is a forum for the users of the CERN Tape Archive (CTA) not general Linux help forum.

Should you have more questions about CTA, feel free to contact us.

Best regards,

Vladimir Bahyl
CERN