HANA does not start: "serviceType mismatch"

[christianhansen]

Hi all,

When I try to start HANA (rev. 85) the nameserver_alert trace file says  "serviceType mismatch" and the daemon trace file says
"

Failed to retrieve HA role for current host from topology, cannot proceed with startup sequence.

Please make sure to properly add additional hosts using addHost/hdbnsutil

"

My system is however single node. It was being used as secondary in a system replication but I deregistered that.

I tried everything I can think of (even restart the server ) but no success.

Anybody with some idea..?

Thanks!

PS. Here is some more outputs from the nameserver_alert trace file:

...

[5937]{-1}[-1/-1] 2015-01-29 19:03:40.616598 e Basis            TREXNameServer.cpp(02906) : Process exited due to an error via explicit exit call with exit code 1 , no crash dump will be written

[6721]{-1}[-1/-1] 2015-01-29 19:54:47.453600 e PersistenceLayer PersistenceSystem.cpp(00221) : serviceType mismatch, expected 0 but found 3

[6721]{-1}[-1/-1] 2015-01-29 19:54:50.061142 e Logger           LogSegment.cpp(00463) : Found not matching segments prevGUID:7B7BC296-0076-20150129-155700-3540000001 myPrevGUID: 

...

[Former Member · Accepted solution]

Hi Christian, You mentioned it was a secondary HANA instance on a SINGLE HOST SYSTEM before it was deregistered.  Can you please the nameserver.ini file under the below path on the Secondary host under: /hana/shared/SID/global/hdb/custom/config [landscape] id = 76183046-abdc-3w24-d230-0000a1109ae7 master = *NODENAME*:30001 idsr = 5413c002-7632-b615-e100-0000b7609b20 You should not see any entries for the Replication ID if it was properly deregistered from the System Replication. It will confirm that the Replication is not causing the issue

[christianhansen · Accepted solution]

Hi Sunil,

Thanks a lot for following this!

This is the content of my nameserver.ini file on the (previous) secondary:

++++++++++++++++++++++++++++++++++++++++++

[landscape]

id = 5451f869-3202-68c9-e100-00000a4c8e3f

master = dewdfglp00766:30001

worker

active_master = dewdfglp00766:30001

+++++++++++++++++++++++++++++++++++++++++++

I even tried with   "hdbnsutil -sr_cleanup --force"  to be really sure

there is not system replication configured anymore.

Cheers

[Former Member · Accepted solution]

Thank you for checking the entries on the nameserver.ini, it looks clean and does not contain anything related to System Replication. 

Please execute a *hdbnsutil -sr_state* to see if there is any other left overs for the System Replicaton

.     Are you looking to use the current state of the HANA system(secondary) or do we have an option to restore this system with Production Backup? We may have to try the restore if -sr_state does not show any entries for the system replication

[christianhansen · Accepted solution]

Hi Sunil,

hdbnsutil -sr_state gives

+++++++++++++++++++++++++++

checking for active or inactive nameserver ...
nameserver dewdfglp00766:30001 not responding.
nameserver dewdfglp00766:30001 not responding.

System Replication State
~~~~~~~~~~~~~~~~~~~~~~~~

mode: none

done.

+++++++++++++++++++++++++++++

When I try   "Recover the database to its most recent state" there is one backup

that shows green on Available when I "Check Availability". But when I execute

I get the error

++++++++++++++++++++++++++++

Could not recover system <SID>

RECOVER DATA finished with error:

recovery strategy could not be determined,

One or more backups cannot be used for recovery

* Could not start master name server

* Backup  / ...   does not exist

* Backup  / ... does not exist

...

+++++++++++++++++++++++++++++

Cheers

[Former Member · Accepted solution]

sr_state does not fnd any left over entries for System Replication so we should be clean

Can you try taking a new backup from the source HANA system and recover it on the Secondary system? Looks like the current backup does not work

[christianhansen · Accepted solution]

Hi,

Finally this was solved. The command line tool hdbnsutil had one option that I had not tried so far:

     hdbnsutil -initTopology

So together with everything else, like  sr_unregister  and sr_cleanup  on both primary and secondary this fixed the issue somehow (I'm not sure about the details though) and both systems are now up running again and I configured system replication between them again.

Thanks for all help!

Best regards

/Chris

[Former Member · Accepted solution]

Excellent Christian

Thank you for sharing this additional step

[Former Member · Accepted solution]

Dear Christian,

Thanks for the information I faced the same issue in our DR database and managed to initialized the topology with below command.

hdbnsutil -initTopology

Regards,

Shanaka.