[X-Servers] Lost XRaid array, no dhcp

Robert Frank robert.frank at unibas.ch
Thu Dec 10 12:50:49 PST 2009


this morning was a catastrophe:

after a seemingly harmless java upgrade on two servers, one suddenly  
lost the array on the xraid and stopped responding to dhcp requests.
The set-up:
xserver PPC with 10.5.8, connected to xRaid via fibre channel with one  
array on the drives in the right bay (8-14)
xserver Intel with 10.5.8 connected to xRaid via fibre channel with  
one array on the drives in the left bay (1-7)
xRaid with 1.5.1 software, two independent arrays, connected to the  
inter and intranet
All green lights, no error messages in the logs anywhere.

Both servers did the update, then rebooted. The PPC server immediately  
found and mounted the array.
The Intel server no longer sees the array. (No SCSI address, no LUN  
when using the system profiler and inspecting the fibre channel).
The array is there, in good condition and available.
The xRaid was powered up all times.

Tried powering off the raid, then reapplying power, rebooted the  
server(s) - nothing changed, no errors reported. The drive simply has  
no SCSI address assigned to it and thus is invisible to the server.  
Lun masking is off. I am currently conditioning the array.
The only good thing is that I have a good backup ... but it would take  
approx. 3 days to rebuild the array and recover from backup.
Apple suggested moving all drives to the other controller, or  
switching controllers. A technician from a company warned about a  
possible data loss...
Any suggestions greatly appreciated.

The other very annoying thing is dhcp. The Intel server is and OD  
master, which also supplied dhcp. I have two networks: the 'internet'  
and a private intranet with no connection to the 'internet'. The dhcp  
server serves both. The addresses for the machines (all servers) on  
the intranet are entered in the static maps of the dhcp service and  
are all ok. The addresses for all machines on the internet are in the  
ldap database (machines) and were always used just fine by dhcp (even  
after reboots). Now, dhcp simply ignores those entries, but ldap is  
used for other things, amongst them login. Again, I see no errors  
which could be linked to this phenomena. Anybody any ideas?

Many, many thanks in advance for all help.

Departement Informatik   FGB   tel   +41 (0)61 267 14 66
Universität Basel                          fax. +41 (0)61 267 14 61
Robert Frank
Klingelbergstrasse 50                 Robert.Frank at unibas.ch
CH-4056 Basel
Switzerland                                   http://www.informatik.unibas.ch/personen/frank_r.html

More information about the X-Servers mailing list