-
vetal
Hey, is it possible to get device disk name by id/wwn (Example in Linux : /dev/disk/by-id/wwn-0x5000c500a4b222a3) ?
-
vetal
Purpose - to get persistent known device name across controllers and reboots, when device name is changing.
-
denk
-
vetal
denk: In some cases that is not true and device name is not exactly as LUN. Logical Unit id: 0x5002538b7210dfa0, but device name is /dev/dsk/c6t5002538B7210DFA2d0
-
vetal
Question was about to have device name w/o controller number, exactly as wwwn, LUN, serial.
-
vetal
Is it possible to configure ?
-
denk
I'm not sure you can configure it manually, only clean device tree file to reorder nodes
-
vetal
denk: In linux: LU WWN Device Id: 5 000c50 0a4b222a3, and device name is: /dev/disk/by-id/wwn-0x5000c500a4b222a3
-
vetal
I.e. exactly as LU WWN
-
denk
you can find something like that in /devices directory
-
denk
vetal: ls -l /devices/scsi_vhci
-
denk
in theory you can find a corresponded device by reading the soft link in /dev/dsk directory
-
tozhu
hello all, I hit / is full on my smartos after my machine is reboot, I think is some core dump or some large file used / file syste, anyone could help how to find that? and thant I can delete the unused file
-
denk
tozhu: df -h /
-
denk
is it a ramdisk?
-
tozhu
yes, but there are a lot of dataset under /zones
-
denk
zones is a pool
-
tozhu
total: 313.28M used: 309.98M avail: 3.30M
-
tozhu
don’t know which file used the space
-
denk
it could be a "deleted" file, where some process has a reference to a file which does not have a corresponded name in any directory
-
denk
tmpfile() returns an opened file descriptor without a name
-
denk
so, if a process uses it, it's too hard to find it
-
tozhu
denk: thanks for the advice
-
vetal
denk: Ho success, device with Logical Unit id: 0x5002538b7210dfa0, is not in /devices/scsi_vhci
-
vetal
denk: However it has Transport protocol: SAS (SPL-4)
-
vetal
denk: It is in /devices/pci@54,0/pci8086,2030@0/pci1000,30e0@0/iport@ff/disk@w5002538b7210dfa2,0, but again
-
vetal
Instead of be as WWN 5002538b7210dfa0, it has 5002538b7210dfa2
-
vetal
Yeah, it is very close to 5002538b7210dfa0, but is bigger by 2 :)
-
denk
vetal: once I saw such behaviour of the system when sas cables were connected incorrect
-
denk
it was with an external jbod connected to two servers
-
vetal
denk: What do mean by incorrect?
-
denk
connector A must be plugged to a hole B, but it was plugged to hole A
-
denk
as result we saw sas-loop
-
denk
run "fmstat 1", column ev_recv must be near 0
-
vetal
denk: Yes, it is 0
-
denk
-
denk
"Are the cables plugged into the correct cards / the correct ports?"
-
vetal
denk: I don't think that it is that case. Drives are seen, but not exactly with the same WWN. It looks like problem in a driver.
-
denk
-
vetal
denk: From it 'the first port will ALWAYS end with "2" and the second port with "3"' that explains ending with 2 for 5002538b7210dfa2.
-
vetal
denk: However, that does not bring solution in general, because for LSISAS3008, device name is changed and
-
vetal
denk: Device is Logical Unit id: 0x5000cca04d1a6ba8
-
vetal
denk: But device name is /dev/rdsk/c39t5000CCA04D1A6BA9d0 , and after reboot it becomes /dev/rdsk/c38t5000CCA04D1A6BAAd0
-
denk
vetal: I have any good ideas, I said what I saw and knew
-
vetal
denk: Even if to wwn 0x5000cca04d1a6ba8 is added "1" to get c39t5000CCA04D1A6BA9, I dont understand why it becomes ending AA after reboot
-
vetal
i.e. c38t5000CCA04D1A6BAA
-
vetal
And It looks like adding 1 or 2 to wwn depends on SAS controller and how get it in general?
-
vetal
I mean algorithm to get proper device name by disk WWN suitable for all SAS controllers.