Why HDB restart is not recommended way to restart HANA node ?

In this We are going to discuss about how HDB restart works and start up sequence of HANA services.

Start up sequence of services depends on run level VALUE.

For example if run valuve for nameserver is 1 means nameserver will be started first
You can check for the used runlevels by the following SQL query .
select * from m_infile_contents where key = ‘runlevel’

You can find these start up sequence in daemon trace.More info on Startup Sequence – Hana Start up sequence

So what happens with HDB restart ?

The command “HDB restart” does not read the runlevels maintained in the daemon.ini. This causes all services to be started simultaneously irrespective of Runlevels.

The simultaneous start of the services might lead to errors that specific services could not be reached.

For instance, the indexserver might not be able to reach its nameserver in time.

In a HANA system replication scenario, the unavailability of the nameserver during manual restart (using HDB restart) might lead to incorrect restart runtime information. In turn, this may cause the re-initiatlization of the data volume which initiates a new full data shipping from the primary site.

This means you’re no longer able to perform a takeover until a new full data shipping was succesful.

[box type=”note” align=”” class=”” width=””]Do not use “HDB restart” when the primary site is unavailable to ensure the ability to takeover to the secondary site.[/box]

So best way to restart hana is

use “HDB stop”

When HANA is stopped completely, Start using “HDB start”

Leave a Reply