Dear all
I would like to receive some comments concerning following.
I have two AS1000A 5/300 (AA and BB) with RAID Array 450 connected to
common scsibus. OS is 4.0D + patchkit 3. TCR is 1.5 + patches.
450 has four raid5 arrays.
ASE is configured so that normally AA handles two raidsets and BB also
two raidsets. Let's call those aseservices AS and BS.
Now, my DLT is connected to AA. To do backups I relocate BS to AA
(asemgr -m BS AA). Relocation failes because rpc.lockd in BB does not
allow dismounting.
First I tackled problem by simply restarting rpc.lockd via rsh in BB
everytime relocation failed but I kept receiving famous "no space in map"
errors. Now I simply reboot BB via rsh if relocation failes BUT sometimes
BB does not boot because the ipaddress it tries to use is already in use
by AA. Of course it is in via automatic service relocation!
Resetting BB (and nic) helps.
BIG question: What is the way from commandline to reboot machine so that
it resets (same function as >>> init does)?
More info: ASEcluster serves hp/ux 9.01-11.00 workstations, irix 5.3-6.5
workstations and dunix 4.0B-4.0D workstations. And one sun. When I use lsof
to find out what rpc.lockd does the offending lock is ALWAYS from hp/ux
10.20.
Another interesting point out of context:
When service BS is relocated to AA sufficiently long time (several hours)
irix 6.2 and above stop "seeing" BS. But only in context of nfs. Ping
to BS works. When BS is relocated back to BB they (irix) workstations
magically start working again. hpux and dunix don't behave like this.
I suspect something wrong with portmapper but I ain't no guru.
Thank You
Tapani
================================================================
VTT/ENERGY Mail tapani.tiensuu_at_vtt.fi
(Tekniikantie 4 C, ESPOO) Tel. +358-9-4565008
P.O.Box 1604 Fax +358-9-4565000
FIN-02044-VTT, FINLAND
================================================================
Received on Fri Jan 22 1999 - 10:42:26 NZDT