Irix 6.5.27 on SGI Origin 350 Kernel Panic

Irix 6.5.27 on SGI Origin 350 Kernel Panic

Post by adrian.per » Wed, 12 Jul 2006 16:42:09


am running Irix 6.5.27 on two SGI Origin 350 's as master and clone
with 4 United Digital Raids, of which are separated into 4 x 1.8TB of
which are mirrored. Also connected is 2 Overland Neo tape units of
which contain 3 drives in each. The problem we had the other weekend
was that the SGI went into a panic and decided to re-boot itself, after
which the Raids went into a revive. We also had temperature warnings in
the analysis.0 log located in /var/adm/crash. Just before the panic
there was a fibre channel error of which was:

Jul 8 18:32:57 6A:uk-pk-sgi-01 unix: qlfc23:
50060b00002e3c76/lun0/c23p1ffffffffe: command timeout: 0x11, notified
demon to abort iocb, port id 0x10a00
Jul 8 18:32:57 6A:uk-pk-sgi-01 unix: qlfc23:
50060b00002e3c76/lun0/c23p1ffffffffe: command timeout: abort of iocb
failed, port id 0x10a00, status 0x4005
Jul 8 18:33:22 6A:uk-pk-sgi-01 unix: qlfc23:
50060b00002e3c76/lun0/c23p1ffffffffe: command timeout: hba failed to
abort iocb, port id 0x10a00, request 0xa800000052824900, cookie

I have managed to find out that this path is connected to the second
Overland Neo tape unit and drive 2. We were having problems with this
drive during the day and prior to the event the drive was taken offline
and placed back online again to try and eliminate the retsore errors we
were getting from that drive.

Here is the rest of the core summary log:

Jul 8 18:33:22 6A:uk-pk-sgi-01 unix: qlfc23: Generating controller
dump and restarting controller.
Jul 8 18:33:47 6A:uk-pk-sgi-01 unix: qlfc23: Firmware version: 3.3.6:
TP.
Jul 8 18:33:47 6A:uk-pk-sgi-01 unix: qlfc23: MBOX: link mode is up.
Jul 8 18:40:57 6F:uk-pk-sgi-01 syslogd: restart
Jul 8 18:40:57 2A:uk-pk-sgi-01 unix: r/6)
Jul 8 18:40:57 5A:uk-pk-sgi-01 unix: NOTICE: Ending XFS recovery for
filesystem: /Raid_PK_4 (/hw/major/268/minor/1509)
Jul 8 18:40:57 5A:uk-pk-sgi-01 unix: NOTICE: Ending XFS recovery for
filesystem: /Raid_PK_3 (/hw/major/268/minor/1495)
Jul 8 18:40:57 5A:uk-pk-sgi-01 unix: NOTICE: Ending XFS recovery for
filesystem: /Raid_PK_1 (/hw/major/268/minor/1499)
Jul 8 18:40:57 5A:uk-pk-sgi-01 unix: NOTICE: Ending XFS recovery for
filesystem: /Raid_PK_2 (/hw/major/268/minor/1488)
Jul 8 18:41:01 5A:uk-pk-sgi-01 unix: NOTICE: tg1: Link up: 1000 Mbps,
FULL duplex, flow_ctrl is OFF
Jul 8 18:41:05 5A:uk-pk-sgi-01 unix: NOTICE: tg0: Link up: 1000 Mbps,
FULL duplex, flow_ctrl is ON
Jul 8 18:41:05 3D:uk-pk-sgi-01 portmap: cannot open tcp6 socket:
Address family not supported by protocol family
Jul 8 18:41:05 3D:uk-pk-sgi-01 portmap: cannot open udp6 socket:
Address family not supported by protocol family
Jul 8 18:41:05 3D:uk-pk-sgi-01 nsd[222]: init (nis): yp is not enabled
by chkconfig
Jul 8 18:41:05 3D:uk-pk-sgi-01 nsd[222]: Dynamic method initialization
failed: nis
Jul 8 18:41:06 3D:uk-pk-sgi-01 rpc.statd[263]: svc_tli_create: could
not open connection for tcp6
Jul 8 18:41:06 3D:uk-pk-sgi-01 rpc.statd[263]: svc_tli_create: could
not open connection for udp6
Jul 8 18:41:06 3A:uk-pk-sgi-01 network: WARNING: autofs: autofs:
/hosts mounted
Jul 8 18:41:06 3A:uk-pk-sgi-01 network: WARNING: autofs: autofs: no
unmounts
Jul 8 18:41:06 5D:uk-pk-sgi-01 inetd[296]: IPv6 service for ftp (and
other servers) ignored - host has no IPv6 addresses
Jul 8 18:41:06 6D:uk-pk-sgi-01 mkpd[395]: daemon started ...
Jul 8 18:41:06 6D:uk-pk-sgi-01 mkpd[395]: partitioning not enabled;
daemon exiting.
Jul 8 18:41:07
 
 
 

1. Irix 6.5.27 Hangs on Octane...and Any Irix Users in London?

2. How to get Secure NFS & Kerberos running in IRIX 6.5.27?

I'm currently struggling with getting Secure NFS & Kerberos running in
a nice and integrated way on a SGI IRIX 6.5.27 system but I'm running
into a couple of annoying misfeatures:

I have /etc/krb5.conf and /etc/krb5/krb5.keytab set up and I can
authenticate against the KDC correctly.

However, I'm not getting the SSH daemon to authenticate against
Kerberos - so I'm wonder what's the way one is supposed to do this
on IRIX systems? My first thought was that one should use PAM but the
pam_krb5 man page warns that it uses it's own Kerberos libraries and
thus doesn't work together with the RPCSEC_GSS system that Secure NFS
use... Compile your own SSH daemon?

How have _you_ folks solved this issue?

Another problem is that even though autofs successfully mounts the
filesystems in question and I can access them (after performing a "kinit"
manually) things only work to a certain degree - apparently "vi" has
problems writing to files - it only creates an empty file and warns
that it couldn't write any characters to the file!

However, if I type "cat >FILENAME" and enter the same information as
I tried in "vi" it works! - Any idea what could be the cause?

No errors in the NFS servers log files.

The NFS Server is a Sun Solaris 9 system.

- Peter
--
--
Peter Eriksson < XXXX@XXXXX.COM > Phone: +46 13 28 2786
Computer Systems Manager/BOFH Cell/GSM: +46 705 18 2786
Physics Department, Linking University Room: Building F, F203

3. Installation of IRIX 6.5.27 from a Powermac cdrom.....can I do it?

4. Irix 6.5.27 - how to obtain ?

5. If you admin IRIX 6.5.21 to 6.5.27...

6. Harpoon3 DB2000 6.5.27 is now available

7. Which Quicktime-version support in 6.5.27?

8. CDROM desktop icon gone missing. 6.5.27

9. Stock IA64 kernel on SGI Altix 350

10. Bug#325117: NFS client problem with kernel 2.6 and SGI IRIX 6.5

11. NFS client problem with kernel 2.6 and SGI IRIX 6.5

12. WTB: Origin 350 system

13. SGI terminates MIPS/IRIX systems (was: SGI BAnkrupt)

14. FS: SGI Origins, Onyxs and SGI Disk Cabinets

15. WTB: SGI Origin 2000/SGI 2100 Deskside (Germany)