[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
SUMMARY: ASE Cluster as Mail server - IMAP + NFS problem
Problem:
> A two node ASE cluster ( Alphaservers 800 5/500 with 2Gb )
> with /var/spool/mail on a shared RA 3000 (raid 5) as an ADVFS filesystem.
> Using an ASE NFS service for serving the mail spool.
>
> After running for a while the system hangs & a crash dump shows:
>
> message buffer full of:
> clntkudp_create: socket bind problem, 48pmap-kgetport: can't create RPC handler.
>
or
> clntkudp_create: socket bind problem: 48clget: null client
Solution:
A couple of replies - basic advice is to move to an ASE Disk service instead of
an NFS service if possible.
From: John.Croft@xxxxxxxxxx
1. ensure your NFS is up to date
back out of pk1
loaded OSFBINOBJECT440 (optional sw)
then re-applied pk1 to get latest NFS code
2. loaded DEC US supplied custom patch for lower level FS code
(yes - I'm quite certain where it was)
our job number was (Australian) Q75043, say you've got similar problems
and want the JCU patch
3. make the ASE service a DISK service only
directly mount it at /var/spool/mail
you may need to re-compile imapd for the better locking real disks give
4. set the both members to default to a 'nullclient sendmail.cf'
eg: /sbin/rc3.d/S99local_nullclient
and also don't start sendmail in /sbin/rc3.d
eg: cd /sbin/rc3.d ; mv S40sendmail no.S40sendmail
your ase start script should kill sendmail, install the smarthost
sendmail.cf and start a sendmail daemon to accept connections only on the member
who owns the disks - the one owning the disks also modifies the /etc/inetd.conf
so that it's inetd spawns off IMAP and POP daemons for requests on the
'ase-mail' service's ip
Thanks to all,
Kevin
-----------------------------------------------------------
Kevin Worvill phone : +44-1603-592392
IT and Computing Services fax : +44-1603-593467
University of East Anglia email : K.Worvill@xxxxxxxxx
Norwich, NR4 7TJ, UK
-----------------------------------------------------------