Language:
switch to room list switch to menu My folders
Go to page: First ... 40 41 42 43 [44] 45 46 47 48 ... Last
[#] Sat Jan 09 2010 22:12:30 EST from Ford II @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

he'll come back in 6 months with hiv and no wife.

[#] Wed Jan 27 2010 11:48:23 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


Hey Linux people...

Does anyone know how the /dev/disk/by-id/* names are calculated?

I'm particularly interested in knowing whether there are any circumstances under which the ID would change, particularly if the disk is re-formatted, re-partitioned, or molested in any other way?

I have a project going right now in which a pair of Linux machines are bringing in Fibre Channel LUN's and then re-exporting them to a farm of VMware boxen via iSCSI. (Yes, I know, the VMware boxen should have Fibre Channel cards in them, but they don't, and there's no budget yet, so this is what I've got to work with.) I like what I'm seeing in /dev/disk/by-id because it appears to be agnostic to the channel path, which is important because each machine has two channels back to the storage.

[#] Wed Jan 27 2010 11:58:37 EST from skpacman @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

If my memory serves me correctly, they're calculated in the order they were identified by their hardware's id built into whatever disk drive it may be. (like windows drive letters) and saved as such, so even if you disconnect disc01 (hardware id 0001) and leave it disconnected for a long time, as long as that configuration is saved then when you plug it back in it should identify it as disc01 and no other hardware id can overtake that...   but i may be wrong...  i'm not entirely 100% sure on that matter.



[#] Wed Jan 27 2010 14:49:45 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

That would imply that they are indeed persistent. I'm looking around on different machines to see what they do. Here's what my desktop shows:

lrwxrwxrwx 1 root root 9 2010-01-22 10:06 scsi-SATA_ST340014AS_5MQ27N97 -> ../../sda

The above was clearly generated from the device's model and serial number.
Similarly, the server we're on now shows something similar:

lrwxrwxrwx 1 root root 9 Jan 19 11:07 scsi-SIBM_DNES18Y_CLAR18_AKG7F770 -> ../../sda
lrwxrwxrwx 1 root root 9 Jan 19 11:07 scsi-SIBM_DNES18Y_CLAR18_AKG78604 -> ../../sdb
lrwxrwxrwx 1 root root 9 Jan 19 11:07 scsi-SSEAGATE_ST336607LC_3JA6MG87 -> ../../sdc
lrwxrwxrwx 1 root root 9 Jan 19 11:07 scsi-SSEAGATE_ST336607LC_3JA6MH9X -> ../../sdd
lrwxrwxrwx 1 root root 9 Jan 19 11:07 scsi-SSEAGATE_ST336607LC_3JA6MN6V -> ../../sde

Here's what I see from within a VMware virtual machine:

lrwxrwxrwx 1 root root 9 Sep 25 09:06 ata-VMware_Virtual_IDE_CDROM_Drive_10000000000000000001 -> ../../hdc

And finally, our machine with the SAN LUN's mapped to it (interestingly, the direct-attached drives are not showing up in this list) ...

lrwxrwxrwx 1 root root 9 Jan 27 06:47 scsi-3600508b300939cb0a97f080e2c03000d -> ../../sda
lrwxrwxrwx 1 root root 9 Jan 27 06:47 scsi-3600508b300939cb0a97f180efb77000d -> ../../sdb
lrwxrwxrwx 1 root root 9 Jan 27 06:47 scsi-3600508b300939cb0a97f280e4ef3000e -> ../../sdc

It should be noted that sometimes we see these as sdd/sde/sdf instead of sda/sdb/sdc because of multi-path topology. And that means it *is* recognizing these as the same disks presented over two different paths, because it's creating three entries, not six.

Ok, so I think I'm optimistic enough now to go ahead and run with the assumption that the ID's aren't going to change. We'll see what happens after I hand these LUN's off to VMware for partitioning and file systems.

[#] Thu Jan 28 2010 12:51:06 EST from Freakdog @ Dog Pound BBS II

[Reply] [ReplyQuoted] [Headers] [Print]

Assuming that the SCSI IDs remain constant, then the disk names should remain constant, as well.



[#] Thu Jan 28 2010 15:19:37 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

That goes against the idea of using /dev/disk/by-id in the first place, which is that the name stays the same even if the path changes. I am seeing the same disk ID's across four paths and two different hosts that share these volumes.

By the way, after partitioning and formatting the volumes, the ID's have not changed.

[#] Thu Jan 28 2010 15:30:35 EST from skpacman @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

By the way, after partitioning and formatting the volumes, the ID's have not changed.

 


So that's good news! :)



[#] Thu Jan 28 2010 22:19:09 EST from Ford II @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

other hardware id can overtake that...   but i may be wrong... 
i'm not entirely 100% sure on that matter.

Definetly not. I shuffled drives, the point of the ID is to find the drive whereever it is physically.
It's probably something dumped in the MBR if Ihad to guess. I think I read about it when I was having all my raid problems, but I don't remember too well because I wasn't concentrating on the id's as much as finding ANY way to find the drive.

[#] Thu Jan 28 2010 22:20:16 EST from Ford II @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

Ok, so I think I'm optimistic enough now to go ahead and run with the

assumption that the ID's aren't going to change. We'll see what

I'm pretty sure the id is burned on to the drive. I don't remember if reparitioning killed it or not, but it's a pretty low leve id.

[#] Thu Jan 28 2010 22:22:46 EST from Ford II @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

By the way, after partitioning and formatting the volumes, the ID's

have not changed.

yeah, then check the mbr.
oh here...
Alternatively, you can use the udev by-id specification (look
in /dev/disk/by-id). The ID paths are usually pretty long
and less meaningful, but they are device specific and won't
change as a result of hardware changes. It's probably best to
use a disk label as above. However, for vfat/fat file systems
disk labels are not available so the by-id specification
should be used.\

[#] Thu Jan 28 2010 23:59:56 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

Filesystem labels are definitely more reliable, and should be used if you're looking to mount a filesystem regardless of its location.

But in this case I'm not looking for a filesystem; I'm looking for a disk volume. The server is acting as a fibre channel to iSCSI bridge, and it's responsible for passing those LUN's through regardless of what filesystems are on it.

[#] Sun Feb 14 2010 21:55:10 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

Not that this is the first thing on everyone's minds right now, but in case anyone is looking for an easy answer to the question "What's a good webcam that works with Linux?" ... I picked up a couple of Logitech C200 cams at Target this weekend for $20 each. They work flawlessly with Linux, have built-in microphones (which appear as USB audio devices) and have a nice good picture quality, even in low light conditions.

I got to videoconference with my kids tonight from a hundred miles away.
When they come home they'll be able to videoconference with their grandmother (at whose home they are visiting this week). It's pretty cool. Yeah, I'm probably the last person on the planet to get on this particular bandwagom, but better late than never. :)

[#] Tue Feb 16 2010 11:46:59 EST from davew @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

 


Yeah, I'm probably the last person on the planet to get on this particular bandwagom, but better late than never. :)

No you're not. I'm still not remotely interested in having a camera on my computer.



[#] Tue Feb 16 2010 11:59:19 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

It's a little more desirable if you have distant friends and family.

[#] Tue Feb 16 2010 12:04:22 EST from fleeb @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


*installs a remote interesting camera on davew's computer*

[#] Tue Feb 16 2010 18:16:47 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

We're going to make davew more interesting, remotely?

[#] Tue Feb 16 2010 22:00:29 EST from Animal @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

This is exactly why I keep a piece of black tape over the webcam on my
computer.

[#] Wed Feb 17 2010 00:00:08 EST from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

You don't have the LED that lights when the camera is activated?

[#] Thu Feb 18 2010 00:50:13 EST from Animal @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

I do, but by the time the green light comes on it's already too late.

[#] Thu Feb 18 2010 09:55:45 EST from Ford II @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

They're coming to get me, but I can't get in front of it.

Go to page: First ... 40 41 42 43 [44] 45 46 47 48 ... Last