I have two HPe tape drives and only one of them is working.
Best I can tell for some reason the OS (RHEL 7.5) seems to have assigned them both the same ID_SERIAL and seems to be mapping both libraries to /dev/tape/by-id/scsi-20404040404040404

Has anyone seen this before? Is there a way to change the ID_SERIAL manually?


#udevadm info --query=all --name /dev/sg16
P: /devices/pci0000:20/0000:20:02.2/0000:24:00.1/host5/rport-5:0-4/target5:0:4/5:0:4:1/scsi_generic/sg16
N: sg16
S: NW_TAPE_1
S: tape/by-id/scsi-20404040404040404 <------ HERE
E: DEVLINKS=/dev/NW_TAPE_1 /dev/tape/by-id/scsi-20404040404040404 <------ HERE
E: DEVNAME=/dev/sg16
E: DEVPATH=/devices/pci0000:20/0000:20:02.2/0000:24:00.1/host5/rport-5:0-4/target5:0:4/5:0:4:1/scsi_generic/sg16
E: ID_MODEL=MSL_G3_Series
E: ID_MODEL_ENC=MSL\x20G3\x20Series\x20\x20\x20
E: ID_REVISION=7.10
E: ID_SCSI=1
E: ID_SCSI_SERIAL=DEC129015YB
E: ID_SERIAL=20404040404040404 <------ HERE
E: ID_SERIAL_SHORT=0404040404040404 <------ HERE
E: ID_TYPE=generic
E: ID_VENDOR=HP
E: ID_VENDOR_ENC=HP\x20\x20\x20\x20\x20\x20
E: MAJOR=21
E: MINOR=16
E: SUBSYSTEM=scsi_generic
E: USEC_INITIALIZED=5362809373

udevadm info --query=all --name /dev/sg16
P: /devices/pci0000:20/0000:20:02.2/0000:24:00.1/host5/rport-5:0-1/target5:0:1/5:0:1:1/scsi_generic/sg9
N: sg9
S: NW_TAPE_2
S: tape/by-id/scsi-20404040404040404 <------ HERE
E: DEVLINKS=/dev/NW_TAPE_2 /dev/tape/by-id/scsi-20404040404040404 <------ HERE
E: DEVNAME=/dev/sg9
E: DEVPATH=/devices/pci0000:20/0000:20:02.2/0000:24:00.1/host5/rport-5:0-1/target5:0:1/5:0:1:1/scsi_generic/sg9
E: ID_MODEL=MSL_G3_Series
E: ID_MODEL_ENC=MSL\x20G3\x20Series\x20\x20\x20
E: ID_REVISION=7.10
E: ID_SCSI=1
E: ID_SCSI_SERIAL=DEC1111BLA
E: ID_SERIAL=20404040404040404 <------ HERE
E: ID_SERIAL_SHORT=0404040404040404 <------ HERE
E: ID_TYPE=generic
E: ID_VENDOR=HP
E: ID_VENDOR_ENC=HP\x20\x20\x20\x20\x20\x20
E: MAJOR=21
E: MINOR=9
E: SUBSYSTEM=scsi_generic
E: USEC_INITIALIZED=19868

I ended up writing my own udev rule but it did not work as desired.

Mine just uses the actual Serial Number rather than the conflicting ID_SERIAL.
However the backup software still did not find the missing library. if someone knows how to fix the ID_SERIAL i would prefer to do that.



# persistent storage links: /dev/tape/{by-id,by-path}

ACTION=="remove", GOTO="persistent_storage_tape_end"

# # type 8 devices are "Medium Changers"
SUBSYSTEM=="scsi_generic", SUBSYSTEMS=="scsi", ATTRS{type}=="8", IMPORT{program}="scsi_id --sg-version=3 --export --whitelisted -d $devnode", \
SYMLINK+="tape/by-id/scsi-$env{ID_SCSI_SERIAL}" <------- UPDATED HERE

SUBSYSTEM!="scsi_tape", GOTO="persistent_storage_tape_end"

KERNEL=="st*[0-9]|nst*[0-9]", ATTRS{ieee1394_id}=="?*", ENV{ID_SERIAL}="$attr{ieee1394_id}", ENV{ID_BUS}="ieee1394"
KERNEL=="st*[0-9]|nst*[0-9]", ENV{ID_SERIAL}!="?*", SUBSYSTEMS=="usb", IMPORT{builtin}="usb_id"
KERNEL=="st*[0-9]|nst*[0-9]", ENV{ID_SERIAL}!="?*", SUBSYSTEMS=="scsi", KERNELS=="[0-9]*:*[0-9]", ENV{.BSG_DEV}="$root/bsg/$id"
KERNEL=="st*[0-9]|nst*[0-9]", ENV{ID_SERIAL}!="?*", IMPORT{program}="scsi_id --whitelisted --export --device=$env{.BSG_DEV}", ENV{ID_BUS}="scsi"
KERNEL=="st*[0-9]", ENV{ID_SERIAL}=="?*", SYMLINK+="tape/by-id/$env{ID_BUS}-$env{ID_SERIAL}"
KERNEL=="nst*[0-9]", ENV{ID_SERIAL}=="?*", SYMLINK+="tape/by-id/$env{ID_BUS}-$env{ID_SERIAL}-nst"

# by-path (parent device path)
KERNEL=="st*[0-9]|nst*[0-9]", IMPORT{builtin}="path_id"
KERNEL=="st*[0-9]", ENV{ID_PATH}=="?*", SYMLINK+="tape/by-path/$env{ID_PATH}"
KERNEL=="nst*[0-9]", ENV{ID_PATH}=="?*", SYMLINK+="tape/by-path/$env{ID_PATH}-nst"

LABEL="persistent_storage_tape_end"

Would it possible to reference the ID_SCSI_SERIAL parameter in your udev rule(s)? At least those are unique.

Cheers...

No, I tried that and it did not work as expected.

I called HPe and had replacement libraries sent to me. The new ones all worked properly out of the box. I assume they had a bad batch of libraries in December based on the serial number sequence - all started with S/N DEC1290

Hi,

I am having the same issue with HPE libraries connected to single host.
Do you have any case ID for case logged with HPE for resolution?
It will be very helpful for reference.

HPe was a joke and told me to talk to my Linux Engineer, which is me.
They swore that it had nothing to do with the library and that it was an OS issue.
If you have a warranty just tell them you are getting a robot error code 80 and to sent a replacement library.
Swapping out the library and robot was the only solution

1 members found this post helpful.

Finally I got replacement of all 5 libraries (MSL4048) with new model MSL3040 from HPE.

Thanks for your suggestion