System Administration
cancel
Showing results for 
Search instead for 
Did you mean: 

Samba CIFS core INTERNAL ERROR: Signal 10 in pid

Samba CIFS core INTERNAL ERROR: Signal 10 in pid

We just upgraded to ADS authentication which required CIFS Server upgrade running on Service Guard. Now we are seeing error messages in daemon.log and network errors when accessing the shares. The link to troubleshooting was no help. Mounting the folders by direct or automount does not seem to matter. MSDFS is set to no as this seems to make Windows 7 clients see empty folders if this is set to yes.

smbd[6214]: From: http://www.samba.org/samba/docs/Samba3-HOWTO.pdf
smbd[6214]: [2011/02/11 07:58:49, 0] lib/fault.c:fault_report(45)
smbd[6214]: [2011/02/11 07:58:49, 0] lib/util.c:smb_panic(1632)
smbd[6214]: PANIC (pid 6214): internal error
smbd[6214]: [2011/02/11 07:58:49, 0] lib/util.c:log_stack_trace(1787)
smbd[6214]: unable to produce a stack trace on this platform
smbd[6214]: [2011/02/11 07:58:49, 0] lib/fault.c:dump_core(181)
smbd[6214]: dumping core in /var/opt/samba/cores/smbd

CIFS-Client A.02.02.02 HP CIFS Client
CIFS-Server A.02.04.01 HP CIFS Server (Samba) File and Print Services
SG-NFS-Tool A.11.31.06 Serviceguard NFS Script Templates
SG-Samba-Tool B.05.00 Serviceguard Samba Script Templates

We also upgraded one server's to latest CIFS Server and it did not help the matter.
One thing we did find is that the CIFS Server uprade force us to modify the hasmb.conf used SG-Samba-Tool kit. For an unknown reason the [SN]MBD_PID_FILE=/var/opt/samba/${NETBIOS_NAME}/locks had to be prefixed with [sn]mbd-smb.conf.${NETBIOS_NAME}.pid previously is was merely [sn]mdb.pid.


We also upgraded one server's to latest CIFS Server A.03.01 and it did not help the matter.

It is very surreal using HP-UX approved packages with such a poor outcome.

5 REPLIES
VK2COT
Honored Contributor

Re: Samba CIFS core INTERNAL ERROR: Signal 10 in pid

Hello James,

Signal 10 is related to bus error.

If you got a core, could you analyse it
via gdb:

http://h21007.www2.hp.com/portal/site/dspp/menuitem.863c3e4cbcdc3f3515b49c108973a801/?ciid=29080f1bace021100f1bace02110275d6e10RCRD

By the way, what are your patch levels of
the HP-UX server? It is obviously running
HP-UX 11.31 but what Update?

Cheers,

VK2COT
VK2COT - Dusan Baljevic

Re: Samba CIFS core INTERNAL ERROR: Signal 10 in pid

Thanks for the reply about the debugger. We are not finding anything in /var/opt/samba/$NETBIOS_NAME/logs/cores. The error on the client side says that the network path is no longer available. Sometimes the folder copy proceeds anyway, othertimes an empty folder is created.

We are patched to the SEPT_2010 DCOE level. We tried setting use mmap = no as it was set to yet by default and swat indicated that HP recommends not using it. But we are still seeing this error. Signal.h does indicate this is a SIGBUS error. But does that indicate the problem's source?

Re: Samba CIFS core INTERNAL ERROR: Signal 10 in pid

Here is some additional information. We are also seeing kernel error messages. This is not occuring on every cluster node.
WARNING: svc_run(): "KLM_svc" SVC thread interrupted [0x28b244b]
WARNING: svc_run(): "NFS_svc" SVC thread interrupted [0x28b244f]


We also decided to look at the kernel and found some kernel changes made via swinstall that did not seem to update anything.
2011-02-09 15:36:56 PST root:
kcmodule -K -s ipf=unused

* Future operations will ask whether to update the backup.
* The requested changes have been applied to the currently
running configuration.
NOTE: No change to the module 'ipf' was needed.
Module State Cause Notes
ipf unused auto-loadable, unloadable

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

2011-02-09 15:36:56 PST root:
kcmodule -K -s pfil=unused

* Future operations will ask whether to update the backup.
* The requested changes have been applied to the currently
running configuration.
NOTE: No change to the module 'pfil' was needed.
Module State Cause Notes
pfil unused auto-loadable, unloadable

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

2011-02-09 15:37:32 PST root:
kcmodule -K pppoe=unused

* Future operations will ask whether to update the backup.
* The requested changes have been applied to the currently
running configuration.
NOTE: No change to the module 'pppoe' was needed.
Module State Cause Notes
pppoe unused loadable, unloadable

It seems unlikely that we are uniquely affected by this problem but it is also improbable that similiar hardware and software configurations are not to be found. Surely we are running standard apps etc.

Re: Samba CIFS core INTERNAL ERROR: Signal 10 in pid

Seeing same problems on 11.23 with CIFS-Server A.02.04. Clients can make folders but when creating files they error out. But a file does get created and is visibile upon reconnect. But clients are unable to edit or save data to any file type. Error states the "network path is no longer available".

Re: Samba CIFS core INTERNAL ERROR: Signal 10 in pid

We may have found a configuration problem. We disabled the SG package and did some testing of samba with the packages file system mounted. The options "map share modes = Yes" and "dos filetime resolution = yes" were in the Global section. We always ran testparm -v and saved the output to a file. No diagnostic messages were seen. I have seen these before and hence alway run testparm before invoking a configuration of smb.conf.

Time will tell but the PANIC SIGNAL 10 errors have abated.