ioscan clarification request

[katie1]

Hi ,I am trying to follow the output I suspect 2 failed disks I marked them with the ## sign in the DS2405 disk enclosures (ioscan -fnk output is attached )


os version B.11.23 server RX2620

I would appreciate a clear understanding / summary of what disks tapes dvd rom etc are attached to which controller and on which ports ,


points assigned to all clear answers ...thanks # do feel free to answer in multiple posts if the answer is long, points assigned to all actual replies ..

===================================

below is how I see it,,imho..

2 root disks mirrored on the internal scisi controller

28 disks in total 4 of these enclosures DS2405

on primary ide channel TEAC DV-28E-C

not sure precisely whats on the second ide channel ?
===========================

HP Tachyon XL2 Fiber Channel Mass Storage Adapter 6 disks on port zero

and 6 more disks on port 1

and the same for the other HP Tachyon XL2 Fiber Channel Mass Storage Adapter

How do I know (Remotely)if a controller is embedded in the system is there away to know from the path?

[Torsten.]

What is your question now?

The disk with ID 4 is showing NO_HW (dead).

IMHO there are 2 DS2405 daisy chained via 2 FC HBAs in slots 3 and 4.

Get more information by running STM info tool on item "HP A6255A" - this is the enclosure (bus) controller card.

[katie1]

so the enclosures shows twice...

let me know what part of the question is unclear so i can clarify if need be,

KT1

[Matti_Kurkela]

I understand the question was:

> How do I know (Remotely)if a controller is embedded in the system is there away to know from the path?

The disk that indicates NO_HW has a hardware path 0/2/1/0.8.0.255.0.4.0.

The structure of the path entry tells me this is a FibreChannel disk. A plain SCSI disk would have a shorter entry (only three numbers separated by dots after the last slash).

In general, the structure of the hardware path is such that if a controller device has hardware path x/y/z, then all the devices controlled by it (and ONLY them) have paths formed like x/y/z*.

In other words, the hardware path of a disk's controller will match the beginning of the disk's hardware path string.

Devices on a PCI/PCI-X bus will typically have a hardware path consisting of four numbers separated by slashes. To identify remotely whether a PCI device is embedded or an expansion card, you need to use "olrad -q" or to find the I/O path diagram in your server's User's Guide.

Looking at the listing above the disk entry, the first non-disk is the ext_bus entry 0/2/1/0.8.0.255.0. That is just the FC bus.

The next entry above it is 0/2/1/0.8, the "FCP protocol adapter". That's not it either (this is a software/firmware entity, not really hardware), but we're getting closer...

Above that is entry 0/2/1/0, "HP Tachyon XL2 Fibre Channel Mass Storage Adapter". That's the controller for your disk.

The next step would be to run "olrad -q", or to look at the hardware documentation for your server. If the hardware path 0/2/1/0 belongs to a card slot, then the controller is not embedded. The hardware path will also allow you to identify the correct slot, if you need that information.

(I happen to know there are no embedded FC controllers in a rx2620, so it must be a card. But this is how you could find it out if you didn't know it.)

You might also run "fcmsutil /dev/td0" and get information about the FC controller, including its model number. Looking up that model number would also tell you it isn't an embedded device.

MK

[katie1]

Matti thanks for your clear reply

I would also appreciate a clear understanding / summary of which disks tapes dvd rom etc are attached to which controller and on which ports ,

and breakdown of total amount of disks and what they are connected too

points assigned to all clear answers .# do feel free to answer in #multiple posts if the answer is long, points assigned to all # actual replies ..

..thanks
KT1

[Matti_Kurkela · Accepted solution]

Let's see...

First, the IDE controller:
0/0/2/0: CMD IDE controller

It has two channels, like an IDE controller always has:
0/0/2/0.0 IDE Primary Channel
0/0/2/0.1 IDE Secondary Channel

There's a DVD drive connected to the primary channel:
0/0/2/0.0.0.0 TEAC DV-28-E, DVD drive c0t0d0

The "Initiator" devices on both IDE channels (0/0/2/0.0.7.0 and 0/0/2/0.1.7.0 respectively) are probably virtual devices generated by the IDE driver, designed to make the IDE controller look like a SCSI controller as much as possible.
On regular SCSI interfaces, one SCSI target ID (usually 7) is reserved for the SCSI controller itself. IDE bus has no concept of SCSI target ID, so the driver is probably faking it enable presenting it to applications as a SCSI bus.

On HP-UX hardware, the IDE controller is integrated to the system board and designed for connecting a DVD drive only. I'm not even sure if the secondary IDE channel has an actual cable connector on the system board or not. (In HP-UX hardware, the secondary IDE channel is like the appendix in humans.)

Each IDE channel occupies one controller ID, so the primary channel was controller c0 and the secondary (which is unlikely to be used by anything ever on HP-UX) is controller c1 in this server.


The next controller (c2) is a real SCSI controller:

0/1/1/0 SCSI Ultra320

There are two disks connected to it:
0/1/1/0.0.0 HP 73.4GST373454LC, disk c2t0d0
0/1/1/0.1.0 HP 73.4GST373454LC, disk c2t1d0

These are 2 of the 3 hot-swap disk slots in the system's front panel. The User's Guide says this controller is integrated to the system board.

The next controller (c3) is again a real SCSI controller, integrated to the system board. If I remember correctly, it controls the 3rd hot-swap disk slot in the front panel, and the SCSI connector in the back.

0/1/1/1 SCSI Ultra320

This controller has no disks (nor anything else) connected to it at the moment. The User's Guide says this is integrated to the system board too.


The next controller is a FibreChannel HBA:
0/2/1/0 HP Tachyon XL2 Fibre Channel Mass Storage Adapter.

It currently uses two controller numbers (c6 and c7) to expand the available range of disk IDs, and could use a lot more if you plugged it to a really big SAN.

It has a lot of disks connected to it:
0/2/1/0.8.0.255.0.0.0 HP 36.4GST336753FC, disk c6t0d0
0/2/1/0.8.0.255.0.1.0 HP 36.4GST336753FC, disk c6t1d0
0/2/1/0.8.0.255.0.2.0 HP 36.4GST336753FC, disk c6t2d0
0/2/1/0.8.0.255.0.3.0 HP 36.4GST336753FC, disk c6t3d0
0/2/1/0.8.0.255.0.4.0 HP 36.4GST336753FC, disk c6t4d0 (currently at NO_HW = disconnected or failed)
0/2/1/0.8.0.255.0.5.0 HP 36.4GST336753FC, disk c6t5d0
0/2/1/0.8.0.255.0.15.0 HP A6255A (the disk enclosure controller, accessible as /dev/rscsi/c6t15d0)

0/2/1/0.8.0.255.1.0.0 HP 36.4GST336753FC, disk c7t0d0
0/2/1/0.8.0.255.1.1.0 HP 36.4GST336753FC, disk c7t1d0
0/2/1/0.8.0.255.1.2.0 HP 36.4GST336753FC, disk c7t2d0
0/2/1/0.8.0.255.1.3.0 HP 36.4GST336753FC, disk c7t3d0
0/2/1/0.8.0.255.1.4.0 HP 36.4GST336753FC, disk c7t4d0
0/2/1/0.8.0.255.1.5.0 HP 36.4GST336753FC, disk c7t5d0
0/2/1/0.8.0.255.1.15.0 HP A6255A (the disk enclosure controller, accessible as /dev/rscsi/c7t15d0)

The disk enclosure controller is a device through which the diagnostics subsystem can monitor things like enclosure's fans and temperatures.

I would have to look up HP model number A6255A to know whether there are 2 separate disk enclosures in a FC loop, or just one enclosure with dual controllers. Both are possible interpretations.


Then there's again a real SCSI controller. It is a dual-channel card, so it provides 2 separate SCSI buses. It has controller numbers c4 and c5, one for each bus.

0/3/1/0 SCSI C1010 Ultra160 Wide LVD A6829-60101
0/3/1/1 SCSI C1010 Ultra160 Wide LVD A6829-60101

On each bus of the card, there's just a single tape drive:
0/3/1/0.2.0 HP C7438A, tape drive c4t2d0* or /dev/rmt/0m*

0/3/1/1.2.0 HP Ultrium 2-SCSI, tape drive c5t2d0* or /dev/rmt/1m*

The last real controller is another FC HBA:
0/5/1/0 HP Tachyon XL2 Fibre Channel Mass Storage Adapter

Like the first HBA, this one uses two controller numbers: c8 and c9.

NOTE: as this FC HBA sees exactly the same configuration of disks as the previous one *and* both HBAs see one failed disk in the same FC hardware path (.../0.8.0.255.0.4.0), this HBA is very likely seeing the same disks as the previous one. This is multipathing, and is commonly done on FibreChannel to provide fault tolerance.

In HP-UX 11.23, the extra paths can and should be configured as "alternate paths" in the LVM VG configuration. (The optional VxVM also has a way to deal with multipathing, but frankly I'm not too familiar with it.)

0/5/1/0.8.0.255.0.0.0 HP 36.4GST336753FC, disk c8t0d0
0/5/1/0.8.0.255.0.1.0 HP 36.4GST336753FC, disk c8t1d0
0/5/1/0.8.0.255.0.2.0 HP 36.4GST336753FC, disk c8t2d0
0/5/1/0.8.0.255.0.3.0 HP 36.4GST336753FC, disk c8t3d0
0/5/1/0.8.0.255.0.4.0 HP 36.4GST336753FC, disk c8t4d0 (currently at NO_HW = disconnected or failed)
0/5/1/0.8.0.255.0.5.0 HP 36.4GST336753FC, disk c8t5d0
0/5/1/0.8.0.255.0.15.0 HP A6255A (the disk enclosure controller, accessible as /dev/rscsi/c8t15d0)

0/5/1/0.8.0.255.1.0.0 HP 36.4GST336753FC, disk c9t0d0
0/5/1/0.8.0.255.1.1.0 HP 36.4GST336753FC, disk c9t1d0
0/5/1/0.8.0.255.1.2.0 HP 36.4GST336753FC, disk c9t2d0
0/5/1/0.8.0.255.1.3.0 HP 36.4GST336753FC, disk c9t3d0
0/5/1/0.8.0.255.1.4.0 HP 36.4GST336753FC, disk c9t4d0
0/5/1/0.8.0.255.1.5.0 HP 36.4GST336753FC, disk c9t5d0
0/5/1/0.8.0.255.1.15.0 HP A6255A (the disk enclosure controller, accessible as /dev/rscsi/c9t15d0)

The final controller is a virtual one, for USB storage devices. If you plugged in an USB DVD-ROM or another supported USB storage device, its hardware path would be prefixed with this "controller":

255/1 USB Mass Storage

MK

[katie1]

Thanks Matti for a fantastic explanation;;

you certainly did clear up the queries I had especially these 3 below I was looking through the ioscan but could not see logic in these,but now it is all clear...



"
1)The "Initiator" devices on both IDE channels (0/0/2/0.0.7.0 and 0/0/2/0.1.7.0 respectively) are probably virtual devices generated by the IDE driver, designed to make the IDE controller look like a SCSI controller as much as possible.
On regular SCSI interfaces, one SCSI target ID (usually 7) is reserved for the SCSI controller itself. IDE bus has no concept of SCSI target ID, so the driver is probably faking it enable presenting it to applications as a SCSI bus.



2)NOTE: as this FC HBA sees exactly the same configuration of disks as the previous one *and* both HBAs see one failed disk in the same FC hardware path (.../0.8.0.255.0.4.0), this HBA is very likely seeing the same disks as the previous one. This is multipathing, and is commonly done on FibreChannel to provide fault tolerance.



3)In HP-UX 11.23, the extra paths can and should be configured as "alternate paths" in the LVM VG configuration. (The optional VxVM also has a way to deal with multipathing, but frankly I'm not too familiar with it.)

thanks Matti

KT1

just one query left ##

so there are 2 DS2405 and each has 6 disks in total with one in no HW state is this correct ?

[Torsten.]

@Matti: "...The next step would be to run "olrad -q"..."

Not on this system, HBAs are not hot swap here.




>> so there are 2 DS2405 and each has 6 disks in total with one in no HW state is this correct ?


Let's have a look at the path:

disk 3 0/2/1/0.8.0.255.0.0.0 sdisk CLAIMED DEVICE HP 36.4GST336753FC
/dev/dsk/c6t0d0 /dev/rdsk/c6t0d0


and


disk 4 0/2/1/0.8.0.255.1.0.0 sdisk CLAIMED DEVICE HP 36.4GST336753FC
/dev/dsk/c7t0d0 /dev/rdsk/c7t0d0



the "1" in the part 255.>>1<<.0.0//255.0.0.0 indicates the setting of the "wheel" switch (the enclosure ID) of the controller, on one enclosure it is set to "0", on the other enclosure it is set to "1".

So there are 2 enclosures behind each of your 2 FC HBAs - daisy chaned, dual pathed.

[katie1]

Thanks Torsten
for clarifying the rest ,

# yet i am a little unclear
in your first post you wrote

"IMHO there are 2 DS2405 daisy chained via 2 FC HBAs in slots 3 and 4."


In this post post you wrote

So there are 2 DS2405 enclosures behind each of your 2 FC HBAs - daisy chaned, dual pathed.


I do understand the ds2405 disks are multipathed ..


So Is it correct that are there 2 ds2405 with 6 disks in each and they are multipathed ..

thanks Torsten
Kt1

[Torsten.]

You are right, the connection is from the server to the first DS2405, to the second DS2405 and back to the server, hence you see all the disks twice (over each HBA).

[katie1]

thanks folks for your clear explanations .
regards
KT1