The bug is in liborchestrator::perform_slim_install.c
It initializes the list of shared filesystems (to be later on
passed to libti) as:
static char *zfs_shared_fs_names[ZFS_SHARED_FS_NUM] =
{"/export", "/export/home", zfs_shared_user_login};
Note the last element is a pointer to a buffer. The problem
is that the buffer is mishandled when a user specifies an ""
empty string for the username in the sc_manifest file:
<property_group name="ai" type="application">
<propval name="username" type="astring" value=""/>
<propval name="userpass" type="astring" value=""/>
The buggy piece of code is in om_perform_install(). When
the username is "", (its not NULL), and it just uses that to build
the user homedir string as "/export/home/%s". This obviously
results in "/export/home/" when username is "", and hence we
run into this bug.
The WORKAROUND is to just completely delete the username
and userpass lines from the SC manifest if you don't want a
user created.
thanks,
-ethan
Ethan Quach wrote:
> Emma,
>
> I just rebooted the system and ran in debug mode. The error
> seems to be that shared filesystem list being passed to be_init()
> contains a duplicate entry. The snippet from the debug log file:
>
> <TIMM_I May 4 23:13:26> Target type to be created: BE
> <TIBEM_I May 4 23:13:26> 1 filesystems will be created
> <TIBEM_I May 4 23:13:26> /
> <TIBEM_I May 4 23:13:26> 3 shared filesystems will be created
> <TIBEM_I May 4 23:13:26> /export
> <TIBEM_I May 4 23:13:26> /export/home
> <TIBEM_I May 4 23:13:26> /export/home/
> <TIBEM_I May 4 23:13:26> bem cmd: /usr/sbin/zpool list rpool 2>&1
> 1>/dev/null
> <TIBEM_E May 4 23:13:29> be_init() failed with error code 4044
>
>
> Notice there's a "/export/home", and a "/export/home/"
> That last "/export/home/" entry is what's causing be_init()
> to fail.
>
> I'll look at why libti is passing in that third entry ....
>
>
> Meanwhile, can you please file a bug on this.
>
>
> thanks,
> -ethan
>
>
>
> Ethan Quach wrote:
>> Emma,
>>
>> Could I just get access to the system for a few minutes?
>>
>>
>> -ethan
>>
>>
>> Emma Lee wrote:
>>> Hi Ethan,
>>>
>>> Here is output from one of system. Do you need any other info or
>>> access to the systm ? Thank you for your help.
>>>
>>> Emma
>>>
>>> sca-maramba-8-sp> console -f
>>> Enter #. to return to ALOM.
>>>
>>> opensolaris console login: root
>>> Password:
>>> Login incorrect
>>> opensolaris console login: root
>>> Password:
>>> Last login: Fri May 1 21:01:48 on console
>>> May 4 14:19:42 opensolaris login: ROOT LOGIN /dev/console
>>> Sun Microsystems Inc. SunOS 5.11 snv_111a November 2008
>>> root at opensolaris:~# zpool list
>>> NAME SIZE USED AVAIL CAP HEALTH ALTROOT
>>> rpool 136G 15.9G 120G 11% ONLINE -
>>> root at opensolaris:~#
>>>
>>> root at opensolaris:/tmp# cat install_log
>>> <AI May 2 03:57:33> /tmp/ai_combined_manifest.xml is a valid manifest
>>> <AI May 2 03:57:33> Auto reboot disabled
>>> <OM May 2 03:57:54> System reports enough physical memory for
>>> installation, swap is optional
>>> <AI May 2 03:57:54> Checking any disks for minimum recommended size
>>> of 12646 MB<AI May 2 03:57:54> Disk c7t0d0 size listed as 140009 MB
>>> <AI May 2 03:57:54> Default disk selected is c7t0d0
>>> <AI May 2 03:57:54> Disk name selected for installation is c7t0d0
>>> <AI May 2 03:57:54> no manifest slice information found
>>> <OM May 2 03:57:54> Timezone setting will be TZ=US/Pacific
>>> <OM May 2 03:57:54> Set timezone
>>> <AI May 2 03:57:54> installation will be performed from
>>> http://ipkg.sfbay.sun.com/dev/ (opensolaris.org)
>>> <AI May 2 03:57:54> installation will be performed from
>>> http://ipkg.sfbay.sun.com/dev/ (opensolaris.org)
>>> <AI May 2 03:57:54> list of packages to be installed is:
>>> <AI May 2 03:57:54> entire
>>> <AI May 2 03:57:54> SUNWcsd
>>> <AI May 2 03:57:54> SUNWcs
>>> <AI May 2 03:57:54> babel_install
>>> <AI May 2 03:57:54> list of packages to be removed is:
>>> <AI May 2 03:57:54> babel_install
>>> <AI May 2 03:57:54> slim_install
>>> <OM May 2 03:57:54> Set zfs root pool device
>>> <OM May 2 03:57:54> creating zpool
>>> <OM May 2 03:57:56> /usr/sbin/zfs get -Hp -o value available rpool
>>> <OM May 2 03:57:56> Creating swap and dump on ZFS volumes
>>> <TIBEM_E May 2 03:58:07> be_init() failed with error code 4044
>>> <OM May 2 03:58:07> Could not create BE target
>>> <OM May 2 03:58:07> TI process failed
>>> <OM May 2 03:58:07> Target instantiation failed exit_val=-1
>>> <AI May 2 03:58:14> om_perform_install failed with error 208
>>> <AI May 2 03:58:14> Auto install failed
>>> root at opensolaris:/tmp#
>>>
>>> Thanks,
>>> Emma
>>>
>>> On 05/04/09 14:11, Ethan Quach wrote:
>>>>
>>>>
>>>> Emma Lee wrote:
>>>>> Hi Ethan,
>>>>>
>>>>> Thank you for helping. I am losting network on the system now, but
>>>>> I will send out zpool list output once I get the network back.
>>>>>
>>>>> Thanks,
>>>>> Emma
>>>>>
>>>>> On 05/04/09 12:23, Ethan Quach wrote:
>>>>>> Emma,
>>>>>>
>>>>>> After the failure, can you do a "zpool list" to make sure the
>>>>>> pool was
>>>>>> created properly. If that succeeds, can you give the output of
>>>>>> "zfs list".
>>>>>>
>>>>>> Do these failures only happen on sparc?
>>>>>>
>>>>>> Can you run this in debug mode to get some additional data:
>>>>>>
>>>>>> sparc: "boot net:dhcp install_debug=enable
>>>>
>>>> Scratch the install_debug part. Enabling debug mode on Sparc
>>>> doesn't seem to have been implemented like this. You have to
>>>> manually edit the install.conf file in the AI image on your server.
>>>> If you have access to this, then edit the <imagedir>/install.conf
>>>> to include a line that says:
>>>>
>>>> install_debug=enable
>>>>
>>>> If you don't have access to do this, getting the data withough
>>>> debug info will also be very useful.
>>>>
>>>> thanks,
>>>> -ethan
>>>>
>>>>
>>>>
>>>>>>
>>>>>> x86: edit the menu entry and add "install_debug=enable" to the
>>>>>> kernel line
>>>>>>
>>>>>>
>>>>>> Attach the /tmp/install_log and /var/svc/log/*install* files, and
>>>>>> whatever shows up on the console too.
>>>>>>
>>>>>>
>>>>>> thanks,
>>>>>> -ethan
>>>>>>
>>>>>>
>>>>>> Emma Lee wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>> We have seen 3 installations with "be_init() failed with error
>>>>>>> code 4044". Will anyone help to resolve this issue ??
>>>>>>>
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Emma
>>>>>>>
>>>>>>> guest domain installation log:
>>>>>>> ====================
>>>>>>> 100%[====================================>] 3,870,720 --.--K/s
>>>>>>>
>>>>>>> 18:59:17 (26.24 MB/s) - `/tmp/solarismisc.zlib' saved
>>>>>>> [3870720/3870720]
>>>>>>>
>>>>>>> --18:59:17--
>>>>>>> http://10.7.81.4:5555/export/aiserver/osol-0906-111a-ai-sparc/install.conf
>>>>>>>
>>>>>>>
>>>>>>> => `/tmp/install.conf'
>>>>>>> Connecting to 10.7.81.4:5555... connected.
>>>>>>> HTTP request sent, awaiting response... 200 OK
>>>>>>> Length: 68 [text/plain]
>>>>>>>
>>>>>>> 100%[====================================>] 68 --.--K/s
>>>>>>>
>>>>>>> 18:59:17 (1.46 MB/s) - `/tmp/install.conf' saved [68/68]
>>>>>>>
>>>>>>> Done mounting automated install image
>>>>>>> Configuring devices.
>>>>>>> Reading ZFS config: done.
>>>>>>> May 1 18:59:33 svc.startd[7]: svc:/platform/sun4u/sckmd:default:
>>>>>>> Method "/lib/svc/method/svc-sckmd" failed with exit status 98.
>>>>>>> May 1 18:59:34 svc.startd[7]: svc:/platform/sun4u/sckmd:default:
>>>>>>> Method "/lib/svc/method/svc-sckmd" failed with exit status 98.
>>>>>>> May 1 18:59:35 svc.startd[7]: svc:/platform/sun4u/sckmd:default:
>>>>>>> Method "/lib/svc/method/svc-sckmd" failed with exit status 98.
>>>>>>> May 1 18:59:35 svc.startd[7]: platform/sun4u/sckmd:default
>>>>>>> failed: transitioned to maintenance (see 'svcs -xv' for details)
>>>>>>> May 1 18:59:36 svc.startd[7]: svc:/platform/sun4u/dscp:default:
>>>>>>> Method "/lib/svc/method/svc-dscp start" failed with exit status 96.
>>>>>>> May 1 18:59:36 svc.startd[7]: platform/sun4u/dscp:default
>>>>>>> misconfigured: transitioned to maintenance (see 'svcs -xv' for
>>>>>>> details)
>>>>>>>
>>>>>>> opensolaris console login: May 1 18:59:45 opensolaris
>>>>>>> mDNSResponder: Local Hostname opensolaris.local already in use;
>>>>>>> will try opensolaris-2.local instead
>>>>>>> May 1 18:59:46 opensolaris mDNSResponder: Local Hostname
>>>>>>> opensolaris-2.local already in use; will try opensolaris-3.local
>>>>>>> instead
>>>>>>> May 1 18:59:46 opensolaris mDNSResponder: Local Hostname
>>>>>>> opensolaris-3.local already in use; will try opensolaris-4.local
>>>>>>> instead
>>>>>>> May 1 18:59:46 opensolaris mDNSResponder: Local Hostname
>>>>>>> opensolaris-4.local already in use; will try opensolaris-5.local
>>>>>>> instead
>>>>>>> Service discovery phase initiated
>>>>>>> Service name to look up: osol-111a-sparc
>>>>>>> Service discovery finished successfully
>>>>>>> Process of obtaining configuration manifest initiated
>>>>>>> Configuration manifest obtained
>>>>>>> Automated Installation started
>>>>>>> The progress of the Automated Installation can be followed by
>>>>>>> viewing the logfile at /tmp/install_log
>>>>>>> Automated Installation failed
>>>>>>> Please refer to /tmp/install_log file for details
>>>>>>> May 1 19:00:22 opensolaris svc.startd[7]:
>>>>>>> application/auto-installer:default failed fatally: transitioned
>>>>>>> to maintenance (see 'svcs -xv' for details)
>>>>>>>
>>>>>>> opensolaris console login: root
>>>>>>> Password:
>>>>>>> May 1 19:00:42 opensolaris login: ROOT LOGIN /dev/console
>>>>>>> Sun Microsystems Inc. SunOS 5.11 snv_111a November 2008
>>>>>>> root at opensolaris:~# cat /tmp/install_log
>>>>>>> <AI May 2 02:00:02> /tmp/ai_combined_manifest.xml is a valid
>>>>>>> manifest
>>>>>>> <AI May 2 02:00:02> Auto reboot disabled
>>>>>>> <TDDM_E May 2 02:00:02> ddm_drive_is_cdrom():
>>>>>>> ioctl(DKIOCREMOVABLE) failed
>>>>>>> <OM May 2 02:00:12> System reports enough physical memory for
>>>>>>> installation, swap is optional
>>>>>>> <AI May 2 02:00:12> Checking any disks for minimum recommended
>>>>>>> size of 12646 MB<AI May 2 02:00:12> Disk c6d0 size listed as
>>>>>>> 16384 MB
>>>>>>> <AI May 2 02:00:12> Default disk selected is c6d0
>>>>>>> <AI May 2 02:00:12> Disk name selected for installation is c6d0
>>>>>>> <AI May 2 02:00:12> no manifest slice information found
>>>>>>> <OM May 2 02:00:12> Timezone setting will be TZ=US/Pacific
>>>>>>> <OM May 2 02:00:12> Set timezone
>>>>>>> <AI May 2 02:00:12> installation will be performed from
>>>>>>> http://ipkg.sfbay.sun.com/dev/ (opensolaris.org)
>>>>>>> <AI May 2 02:00:12> installation will be performed from
>>>>>>> http://ipkg.sfbay.sun.com/dev/ (opensolaris.org)
>>>>>>> <AI May 2 02:00:12> list of packages to be installed is:
>>>>>>> <AI May 2 02:00:12> entire
>>>>>>> <AI May 2 02:00:12> SUNWcsd
>>>>>>> <AI May 2 02:00:12> SUNWcs
>>>>>>> <AI May 2 02:00:12> babel_install
>>>>>>> <AI May 2 02:00:12> list of packages to be removed is:
>>>>>>> <AI May 2 02:00:12> babel_install
>>>>>>> <AI May 2 02:00:12> slim_install
>>>>>>> <OM May 2 02:00:12> Set zfs root pool device
>>>>>>> <OM May 2 02:00:12> creating zpool
>>>>>>> *<OM May 2 02:00:15> /usr/sbin/zfs get -Hp -o value available rpool
>>>>>>> <OM May 2 02:00:15> Creating swap and dump on ZFS volumes
>>>>>>> <TIBEM_E May 2 02:00:18> be_init() failed with error code 4044
>>>>>>> <OM May 2 02:00:18> Could not create BE target
>>>>>>> <OM May 2 02:00:18> TI process failed *
>>>>>>> <OM May 2 02:00:18> Target instantiation failed exit_val=-1
>>>>>>> <AI May 2 02:00:22> om_perform_install failed with error 208
>>>>>>> <AI May 2 02:00:22> Auto install failed
>>>>>>> root at opensolaris:~#
>>>>>>> ------------------------------------------------------------------------
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> caiman-discuss mailing list
>>>>>>> caiman-discuss at opensolaris.org
>>>>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>>>
>>>
>> _______________________________________________
>> caiman-discuss mailing list
>> caiman-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
