Hi Stefan,
On Thu, 26 Jan 2023 16:35:59 +0100, Stefan Kania via samba
<samba at lists.samba.org> wrote:
> I'm having a CTDB-Cluster with two nodes (both Ubuntu wit
> Sernet-packages 4.17.4). Now I want to replace one of the nodes. The
> first step was to bring a new node to the CTDB-Cluster. This time a
> Debian 11 but with the same sernet-packages (4.17.4). Adding the new
> node to /etc/ctdb/nodes at the end of the list. And the virtual IP to
> /etc/ctdb/public_addresses, also at the end of the list. Copy both files
> to the new server. Then mounting the Glusterfs. Configuring the
> recovery-log. Then starting sernet-samba-ctdb. The ctdb is starting and
> the new host is showing "OK" in "ctdb status".
>
> Content of smb.conf
>
> [global]
>
>
> clustering = yes
>
>
> include = registry
>
>
> Then "net conf list" is showing the configuration from registy.
> smbstatus is showing the same connection as any other of the CTDB-Hosts.
>
> Now I enabled the first event 49.winbind with:
>
> ctdb event script enable legacy 49.winbind
>
> the node chrashes with the following messages in /var/log/log.ctdb
>
> ------------------
> 2023-01-26T16:14:30.603370+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> ERROR: wbinfo -p returned error
>
> 2023-01-26T16:14:30.603446+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> could not ping winbindd!
>
> 2023-01-26T16:14:30.603458+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> Ping to winbindd failed
>
> 2023-01-26T16:14:30.603580+01:00 glfs3 ctdbd[15762]: monitor event
> failed - disabling node
>
> 2023-01-26T16:14:30.604273+01:00 glfs3 ctdbd[15762]: Node became
> UNHEALTHY. Ask recovery master to reallocate IPs
>
> 2023-01-26T16:14:31.302861+01:00 glfs3 ctdb-recoverd[15867]: Disabling
> takeover runs for 60 seconds
>
> 2023-01-26T16:14:31.323234+01:00 glfs3 ctdbd[15762]: Release of IP
> 192.168.0.227/24 on interface ens18 node:0
>
> 2023-01-26T16:14:31.409901+01:00 glfs3 ctdb-recoverd[15867]: Reenabling
> takeover runs
>
> 2023-01-26T16:14:35.654416+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> ERROR: wbinfo -p returned error
> ...
> 2023-01-26T16:14:35.654473+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> Ping to winbindd failed
>
> 2023-01-26T16:14:45.711894+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> ERROR: wbinfo -p returned error
>
> 2023-01-26T16:14:45.711944+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> could not ping winbindd!
>
> 2023-01-26T16:14:45.711955+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> Ping to winbindd failed
>
> 2023-01-26T16:15:00.757976+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> ERROR: wbinfo -p returned error
>
> 2023-01-26T16:15:00.758016+01:00 glfs3 ctdb-eventd[15763]: 49.winbind:
> could not ping winbindd!
> ------------------
>
> disabling the event and the node is getting back to "OK"
>
> Now enabling 50.samba and the node is crashing again with the following
> messages:
>
> --------
> 2023-01-26T16:20:31.830034+01:00 glfs3 ctdb-eventd[15763]: 50.samba:
> samba not listening on TCP port 445
> 2023-01-26T16:20:31.830139+01:00 glfs3 ctdbd[15762]: monitor event
> failed - disabling node
> 2023-01-26T16:20:31.830557+01:00 glfs3 ctdbd[15762]: Node became
> UNHEALTHY. Ask recovery master to reallocate IPs
> 2023-01-26T16:20:31.849116+01:00 glfs3 ctdb-recoverd[15867]: Disabling
> takeover runs for 60 seconds
> 2023-01-26T16:20:31.857072+01:00 glfs3 ctdbd[15762]: Release of IP
> 192.168.0.227/24 on interface ens18 node:0
> 2023-01-26T16:20:31.936495+01:00 glfs3 ctdb-recoverd[15867]: Reenabling
> takeover runs
> 2023-01-26T16:20:36.919191+01:00 glfs3 ctdb-eventd[15763]: 50.samba:
> samba not listening on TCP port 445
> 2023-01-26T16:20:47.013031+01:00 glfs3 ctdb-eventd[15763]: 50.samba:
> samba not listening on TCP port 445
> 2023-01-26T16:21:02.104424+01:00 glfs3 ctdb-eventd[15763]: 50.samba:
> samba not listening on TCP port 445
> -----------
>
> Again checking the configuration on all hosts, all the same.
>
>
> /etc/default/sernet-samba is set to "classic"
>
>
> Any help is welcome :-)
winbindd and smbd are only started during the CTDB startup event, so
you have to enable the event scripts before you start CTDB.
That hasn't changed... :-)
peace & happiness,
martin