connect(/var/lib/ctdb/ctdb.socket) failed: Connection refused

[Martin Schwenke]

On Sun, 10 Aug 2014 11:13:24 +0200, steve <steve at steve-ss.com> wrote:

> For now. All sorts of anomalies already. Even though the binaries have
> gone in under /usr/local, the event scripts have ended up in distroland
> under /etc

Not sure what's happening at your end.  For me, on a Debian system with
out-of-the-box ctdb-2.5.3:

  $ find /usr/local/etc -type f 
  /usr/local/etc/ctdb/notify.sh
  /usr/local/etc/ctdb/ctdb-crash-cleanup.sh
  /usr/local/etc/ctdb/nfs-rpc-checks.d/10.statd.check
  /usr/local/etc/ctdb/nfs-rpc-checks.d/50.rquotad.check
  /usr/local/etc/ctdb/nfs-rpc-checks.d/20.nfsd.check
  /usr/local/etc/ctdb/nfs-rpc-checks.d/40.mountd.check
  /usr/local/etc/ctdb/nfs-rpc-checks.d/30.lockd.check
  /usr/local/etc/ctdb/functions
  /usr/local/etc/ctdb/debug_locks.sh
  /usr/local/etc/ctdb/gcore_trace.sh
  /usr/local/etc/ctdb/debug-hung-script.sh
  /usr/local/etc/ctdb/events.d/99.timeout
  /usr/local/etc/ctdb/events.d/11.natgw
  /usr/local/etc/ctdb/events.d/20.multipathd
  /usr/local/etc/ctdb/events.d/60.ganesha
  /usr/local/etc/ctdb/events.d/40.fs_use
  /usr/local/etc/ctdb/events.d/10.interface
  /usr/local/etc/ctdb/events.d/13.per_ip_routing
  /usr/local/etc/ctdb/events.d/91.lvs
  /usr/local/etc/ctdb/events.d/01.reclock
  /usr/local/etc/ctdb/events.d/49.winbind
  /usr/local/etc/ctdb/events.d/60.nfs
  /usr/local/etc/ctdb/events.d/31.clamd
  /usr/local/etc/ctdb/events.d/00.ctdb
  /usr/local/etc/ctdb/events.d/40.vsftpd
  /usr/local/etc/ctdb/events.d/50.samba
  /usr/local/etc/ctdb/events.d/41.httpd
  /usr/local/etc/ctdb/events.d/62.cnfs
  /usr/local/etc/ctdb/events.d/70.iscsi
  /usr/local/etc/ctdb/events.d/11.routing
  /usr/local/etc/ctdb/statd-callout
  /usr/local/etc/sudoers.d/ctdb

I suppose it could be Ubuntu strangeness.

If you set CTDB_BASE somewhere (/etc/profile.d/ctdb.sh ?) before
running ctdbd_wrapper then you could even have
/usr/local/etc/ctdb/ctdbd.conf.

We clearly need to fix the default "make install" so that it just
works.  It seems clear that we tend to test with RPMs...

peace & happiness,
martin