This bug is still valid in opensuse-13.2

Created attachment 633844 [details]
Updated rescan-scsi-bus.sh

Please re-test using the attached rescan-scsi-bus.sh version

OK i tried the new script but nothing changed. My SCSI devices before:

# lsscsi -g
[0:0:0:0]    disk    ATA      ST3160815AS      B     /dev/sda   /dev/sg0 
[1:0:0:0]    disk    ATA      ST3160815AS      D     /dev/sdb   /dev/sg1 
[2:0:0:0]    disk    ATA      ST3160811AS      E     /dev/sdc   /dev/sg2 
[3:0:0:0]    disk    ATA      ST3160815AS      D     /dev/sdd   /dev/sg3 
[6:0:0:0]    disk    ATA      ST3500320NS      SN06  /dev/sde   /dev/sg4 
[7:0:0:0]    disk    ATA      ST3500320NS      SN04  /dev/sdf   /dev/sg5 
[8:0:0:0]    cd/dvd  SONY     DVD RW AD-7260S  1.61  /dev/sr0   /dev/sg6 
[8:0:1:0]    disk    ATA      ST3500418AS      CC38  /dev/sdg   /dev/sg7 
[10:0:2:0]   tape    HP       88780            A674  /dev/st1   /dev/sg10
[10:0:6:0]   tape    EXABYTE  EXB-8500-85Qanx0 0446  /dev/st2   /dev/sg11
[11:0:0:0]   disk    SEAGATE  ST15150N         5055  /dev/sdh   /dev/sg8 
[11:0:3:0]   tape    SONY     SDT-10000        0110  /dev/st0   /dev/sg9

I removed the SONY tape drive on bus 11 at /dev/st0 , /dev/sg9 and added another SEAGATE SCSI disk at SCSI ID 1 on bus 11.

#./rescan-scsi-bus.sh 

Scanning SCSI subsystem for new devices
Scanning host 0 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg0 changed:  device 0 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3160815AS      Rev: B   
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 1 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg1 changed:  device 1 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3160815AS      Rev: D   
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 2 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg2 changed:  device 2 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3160811AS      Rev: E   
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 3 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg3 changed:  device 3 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3160815AS      Rev: D   
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 4 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
Scanning host 5 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
Scanning host 6 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg4 changed:  device 6 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3500320NS      Rev: SN06
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 7 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg5 changed:  device 7 0 0 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3500320NS      Rev: SN04
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 8 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg6 changed:  device 8 0 0 0 ...           
from:CD-ROM                            : 00
to: CD-ROM  : SONY     Model: DVD RW AD-7260S  Rev: 1.61
      Type:   CD-ROM                           ANSI SCSI revision: 05
                

sg7 changed:  device 8 0 1 0 ... 
from:Direct-Access                     : 00
to: Direct-Access      Model: ST3500418AS      Rev: CC38
      Type:   Direct-Access                    ANSI SCSI revision: 05


Scanning host 9 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
Scanning host 10 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg10 changed: device 10 0 2 0 ...          
from:Sequential-Access                 n: 00
to: Sequential-Access  Model: 88780            Rev: A674
      Type:   Sequential-Access                ANSI SCSI revision: 01


sg11 changed: device 10 0 6 0 ...          
from:Sequential-Access                 n: 00
to: Sequential-Access  Model: EXB-8500-85Qanx0 Rev: 0446
      Type:   Sequential-Access                ANSI SCSI revision: 02
                  

Scanning host 11 for  SCSI target IDs  0 1 2 3 4 5 6 7, all LUNs
sg8 changed:  device 11 0 0 0 ...          
from:Direct-Access                     n: 00
to: Direct-Access  TE  Model: ST15150N         Rev: 5055
      Type:   Direct-Access                    ANSI SCSI revision: 02


sg9 changed: INQUIRY failed    ..          
OLD: Host: scsi11 Channel: 00 Id: 03 Lun: 00
      Vendor: SONY     Model: SDT-10000        Rev: 0110
      Type:   Sequential-Access                ANSI SCSI revision: 02
0 new or changed device(s) found.          
0 remapped or resized device(s) found.
0 device(s) removed.


                
harley:/home/markh # lsscsi -g
[0:0:0:0]    disk    ATA      ST3160815AS      B     /dev/sda   /dev/sg0 
[1:0:0:0]    disk    ATA      ST3160815AS      D     /dev/sdb   /dev/sg1 
[2:0:0:0]    disk    ATA      ST3160811AS      E     /dev/sdc   /dev/sg2 
[3:0:0:0]    disk    ATA      ST3160815AS      D     /dev/sdd   /dev/sg3 
[6:0:0:0]    disk    ATA      ST3500320NS      SN06  /dev/sde   /dev/sg4 
[7:0:0:0]    disk    ATA      ST3500320NS      SN04  /dev/sdf   /dev/sg5 
[8:0:0:0]    cd/dvd  SONY     DVD RW AD-7260S  1.61  /dev/sr0   /dev/sg6 
[8:0:1:0]    disk    ATA      ST3500418AS      CC38  /dev/sdg   /dev/sg7 
[10:0:2:0]   tape    HP       88780            A674  /dev/st1   /dev/sg10
[10:0:6:0]   tape    EXABYTE  EXB-8500-85Qanx0 0446  /dev/st2   /dev/sg11
[11:0:0:0]   disk    SEAGATE  ST15150N         5055  /dev/sdh   /dev/sg8 
[11:0:3:0]   tape    SONY     SDT-10000        0110  /dev/st0   /dev/sg9

Could you please run the script with sh -x and attach the logs here, so I can have a look what's actually going wrong.

Thanks

Created attachment 634136 [details]
Output from "sh -x   rescan-scsi-bus.sh"

Had to xz the text attachment because it exceeded the 10240kb attachment file limit

Seems like my error, I've only attached a patched version of the rescan-scsi-bus.sh script, but forgot to update sg_turs as well.

I'll build an updated rpm shortly and notice you once it is done.

Sorry for the inconvenience.

Can you please try the sg_turs from https://build.opensuse.org/package/show/home:morbidrsa:branches:Base:System/sg3_utils

Maybe you'll need to use the rescan-scsi-bus.sh from this ticket.

Thanks

I can't seem to find an rpm to try??

Sorry, somehow the publish flag was cleared...
Should be accessible now

OK, I got and installed them. No change though??

Mark

The problem (with the old sg_turs from sg3_utils) was that it had a return code of 99, which isn't known by the script. This should be fixed with the preliminary sg3_utils package I gave you.

Can you verify it? What's the output of
sg_turs /dev/sg9; echo $?
assuming sg9 still is the SONY tape drive in question.

Have you tried with the rescan-scsi-bus.sh from the rpm or with the one attached to this bug?

With the tape installed 0 is returned. With it powered off, the following is returned.

#sg_turs /dev/sg11; echo $?
test unit ready: transport: Host_status=0x01 [DID_NO_CONNECT]
Driver_status=0x00 [DRIVER_OK, SUGGEST_OK]

Sense category: -1
99

This is using the 2 rpms. 

Note that devices removed were never actually removed from the output of the "lsscsi -g" command even though the script would report it was removed and when reinstalled and the script re-ran it would still show it's original designation via "lsscsi -g". But new devices were reported and always added and would be available for use after running the script. Devices added are my biggest concern.

FYI, the rescan-scsi-bus.sh script from the 12.3 distribution works when run on my 13.2 system. It correctly finds new devices and they are available for use after running that script. Removed devices are still not shown as removed but that is not really important. New devices found are what is critical and that works using the rescan-scsi-bus.sh from the 12.3 dist on my 13.2 system.

Hi,

We've updated sg3_utils to the newest version in Factory [1], can you have a look at it?

I'm sorry that this is more try and error than anything else but I don't know how to solve this issue.

Thanks

[1] https://build.opensuse.org/package/show?project=Base%3ASystem&package=sg3_utils

That does not fix it. As I said:

"FYI, the rescan-scsi-bus.sh script from the 12.3 distribution works when run on my 13.2 system. It correctly finds new devices and they are available for use after running that script. Removed devices are still not shown as removed but that is not really important. New devices found are what is critical and that works using the rescan-scsi-bus.sh from the 12.3 dist on my 13.2 system."

????

Who is the official maintainer of this package and script. I can't find anything on the web as to who to report bugs to. The script from the SuSE-12.3 package works just fine. Someone either just needs to swap it out or ask the "real" owner of this package to fix it.

(In reply to Mark Hounschell from comment #15)
> That does not fix it. As I said:
> 
> "FYI, the rescan-scsi-bus.sh script from the 12.3 distribution works when
> run on my 13.2 system. It correctly finds new devices and they are available
> for use after running that script. Removed devices are still not shown as
> removed but that is not really important. New devices found are what is
> critical and that works using the rescan-scsi-bus.sh from the 12.3 dist on
> my 13.2 system."
> 
> ????

Can you please check if it work with the new script if you call it with 

rescan-scsi-bus.sh -r

?

If not, can you please provide the output of 'bash -x rescan-scsi-bus.sh -r' from both, the old, working one and the new one?

Thanks.

Created attachment 651181 [details]
Output of 'sh -x /usr/bin/rescan-scsi-bus-12.3.sh -r'

Created attachment 651182 [details]
Output of 'sh -x /usr/bin/rescan-scsi-bus.sh -r'

Using the libsgutils2-1_41-2-1.41-64.1.x86_64.rpm and sg3_utils-1.41-64.1.x86_64.rpm packages from factory, the 'rescan-scsi-bus.sh -r' command did not find a new device. I've attached the output of the rescan scripts from the above and the 12.3 script

Have just tried this on Tumbleweed with sg3_utils 1.43 - added a drive, rescan-scsi-bus.sh did not find it.

I'd rather lay this bug to rest; please open another bug if it's still an issue.

This bug is still as valid today as it was when it was filed. Why should I refile it when you all will just let it rot again. Shameful...

Because it has been filed against a (now unsupported) version.
Can you please re-file it against the most recent version?
Thanks.

It is still broken in Leap 42.3. I see no valid reason to refile this bug.

So change the Version: tags of the bug?

This bu has not been FIXED

I have verified that this bug is still a bug on leap 42.3 and also on leap-15 beta-197.1. I can also that the rescan-scsi-bus script from the SuSE-12.3 sg_utils package does in fact work properly on leap-42.3 and leap-15 beta-197.1 

I have also changed the VERSION ID from 42.3 to 15. Lets see this bug is now over 4 years old.

This bug still exists even though the original BUG report was back in 2015. The last rescan-scsi-bus.sh that actually worked was from the 12.3 dist. I have had to keep the version for 12.3 on my system every since because I use it constantly. Please fix this BUG. I would be happy to attach the 12.3 version of this script but find no way to do it but inline here.

Created attachment 852026 [details]
diff between 12.3 and 13.1 versions

obtained with unrpm on
/mounts/schnell/CD-ARCHIVE/openSUSE/12.3/POOL/suse/x86_64/sg3_utils-1.33-5.1.1.x86_64.rpm
/mounts/schnell/CD-ARCHIVE/openSUSE/13.1/POOL/suse/x86_64/sg3_utils-1.36-3.1.2.x86_64.rpm

The only change from comment 31 that could have caused the change in behavior is this, afaics:

-    # We used to always trigger a rescan for FC to update channels and targets
-    # Commented out -- as discussed with Hannes we should rely
-    # on the main loop doing the scan, no need to do it here.
-    #echo "- - -" > /sys/class/scsi_host/host$host/scan 2> /dev/null;
-    #udevadm_settle

One thing that stands out is the removed "udevadm settle". 

Could you locate this code:

> idlist ()
> {
>   local tmpid
>   local newid
>   local oldid
> 
>  oldlist=$(ls /sys/class/scsi_device/ | sed -n "s/${host}:${channel}:\([0-9]*:[0-9]*\)/\1/p" | uniq)
>  # Rescan LUN 0 to check if we found new targets
>  echo "${channel} - -" > /sys/class/scsi_host/host${host}/scan

and put an "udevadm settle" below the last line ('echo "${channel} - -"...') and see if it helps?

correction: put "udevadm_settle" like in the original code (with underscore), not just "udevadm settle".

I've added it. 

# Returns the list of existing targets per host
idlist ()
{
  local tmpid
  local newid
  local oldid

  oldlist=$(ls /sys/class/scsi_device/ | sed -n "s/${host}:${channel}:\([0-9]*:[0-9]*\)/\1/p" | uniq)
  # Rescan LUN 0 to check if we found new targets
  echo "${channel} - -" > /sys/class/scsi_host/host${host}/scan

  udevadm_settle

  newlist=$(ls /sys/class/scsi_device/ | sed -n "s/${host}:${channel}:\([0-9]*:[0-9]*\)/\1/p" | uniq)

It did not help though.

Let me go back and verify which version actually broke it. I went from 12.3 to 13.2. I never used 13.1.

The rescan-scsi-bus.sh from the 13.1 dist also works. It breaks at 13.2. I just now verified this.

That would be good. I think there must be more differences than those in comment 31.

12.3 GA: 1.33
12.3 Update: 1.33 + some patches
13.1: 1.36
13.2: 1.39

However, rescan-scsi-bus.sh may have changed without being reflected in the version number, there's notes like this:

> -------------------------------------------------------------------
> Mon Aug  4 18:13:25 UTC 2014 - spargaonkar@suse.com
> 
> - Replaced rescan-scsi-bush.sh in sg3_utils-1.36.tar.xz

... good we aren't working like that any more ...

Just an FYI, I am running these scripts on 15.3. I have plenty of SCSI devices to play with so anything you want me to try, I can.

This coincides with the introduction of the exisiting_targets logic.

Maybe it's as simple as running 'rescan-scsi-bus.sh -a --hosts=11' ?

Please run the current script with the "-a" flag, and (if it doesn't work) provide a sh -x log file like you did earlier.

Ah, the "-a" flag seems to do the trick. Do you still want the "sh -x" log file?

No. Please confirm that it works without the change from comment 34.

It's true that the addition of "-a" was a change in the semantics of the script.
This was before my time, I don't know why it has been done (I wasn't involved in sg3_utils at that time, 2013). The changelog doesn't say anything, either.

If you're ok, I'm going to close this bug. Apologies that it took so long.

Yes, I have verified that the rescan script in 15.3 works with the "-a" flag as far as discovering newly added devices. 

This script originally discovered new and removed devices with no "-flags". It appears now that in order for it to work as it did originally you must pass not only the "-a" but the "-r" flag also. The "-r" flag removes removed devices from the list displayed by the "lsscsi -g" command.

Yes, you can close this bug now. Thank you.

So "-r" worked for you? 

In some cases you may need `--forceremove` to actually delete the devices from the host.

Thanks for your patience.