[Samba] Unable to "rejoin" existing DC after upgrade (infamous WERR_FILE_NOT_FOUND)

[Rowland Penny]

On 14/04/2023 12:13, Lorenzo Milesi via samba wrote:
> 
>> That all smells quite similar to what I were seeing
>> here before, and what someone else were seeing too,
>> all with 4.17[.4]. This famouse WERR_FILE_NOT_FOUND
>> error too.
> 
> We upgraded one more domain, and in the hope to avoid this error, we remained on 4.16. So we deployed a new 4.16.9 DC, joined, everything is fine.
> Remove the old DC, upgrade OS, install 4.16.10 (which was released in the meantime), join back, everything is fine.
> 
> After "some time", on the second DC (the one implemented for upgrading the first):
> 
> DC=DomainDnsZones,DC=wdc,DC=dom,DC=it
>          Default-First-Site-Name\DC1 via RPC
>                  DSA object GUID: 60b7eeda-0650-4efe-a576-569dfc012847
>                  Last attempt @ Fri Apr 14 11:39:38 2023 CEST failed, result 2 (WERR_FILE_NOT_FOUND)
>                  1 consecutive failure(s).
>                  Last success @ Fri Apr 14 11:39:38 2023 CEST
> 
> 
> The replica has been working fine for one or two hours (didn't constantly monitor, but was checked regularly). Then we moved FSMO roles "back" to dc1, and the error popped up.
> I cannot say it this has been the trigger, but this is the only change we did after the upgrade.
> 
> Now dc1 shows ALL GOOD, while dc2 shows all replication items with errors.
> 
> 
> Is there any way I can determine dc1 being "correct" or not? FSMO roles are all in charge to that.
> 
> thanks again


Can you please run this command on a DC and post the output:

host -t SRV _ldap._tcp.pdc._msdcs.wdc.dom.it

Rowland