EFM Failover issue when Master server restarts

[pcpg]

Hi Ninad,

Can you pls elaborate a bit?

What is happening in RHEL 6 and what is not happening in RHEL 7?

Also, since the behaviour is not consistent, does that not indicate an issue?

Thanks

[nshah]

Hi Tom,

Thank you for your respnose. Automatic failover on graceful reboot may not happend with SuSE Linux. However, This is happening on CentOS/RHEL 6.x

Thanks & Regards,

Ninad Shah

[nshah]

Hi Tom

Also, since the behaviour is not consistent, does that not indicate an issue?

Thanks

---

It may happen that failover may be cancelled when Master is brought back online within alloted time out.

Thanks & Regards,

Ninad Shah

[pcpg]

Hi Ninad,

On this thread, Dhananjay had mentioned that as per last line of the foll. page https://www.enterprisedb.com/docs/en/3.1/edbfm/EDB_Failover_Manager_Guide.1.31.html

standby will not be promoted when stopping an agent. In case of "init 6" or "shutdown" command, believe the agent is considered to have been shutdown gracefully and hence promotion should not happen, isnt it?

But as Tom has reported and as you have also confirmed, automatic failover is happening on graceful reboot. So if the behaviour is correct and expected then is it a documentation bug?

Thanks

[nshah]

Hi pcpg,

Thank you for your response over the ticket.

Automatic reboot upon graceful restart (init 6/reboot) is happening on RHEL/CentOS 6.x platform. However, This is not happening on RHEL/CentOS 7.x or SuSE platforms.

Hope this response may helpful to you. Request you to let us know If the answer satisfies your query.

Thanks & Regards,

Ninad Shah

[pcpg]

Hi Ninad,

No, it does not satisfy. Reason being, manuals say that promotion should not happen whereas when we test, we see that promotion is happening. So which one is correct/expected?

Thanks

[nshah]

Hi pcpg,

Apology for bad explanation in previous response.

Automatic promotion/failover of slave upon graceful restart (init 6/reboot), is happening on RHEL/CentOS 6.x platform. However, This is not happening on RHEL/CentOS 7.x or SuSE platforms.

Hope this may help you. Request you to let us know If that response satisfies your query.

Thanks & Regards,

Ninad Shah

[pcpg]

Unfortunately not Ninad.

Dhananjay mentioned that promotion will not happen.

Manual also says that promotion will not happen.

Tom confirmed that promotion happens on RHEL but not on Suse.

We also tested and promotion happens on RH.

So is the manual incorrect?

Hope i am able to explain the concern.

[chaitalirs]

Hi pcpg,

Apologies for the confusion created. The EFM behaviour will be same on all the operating systems (Supported platforms).

Please if you could provide following details to have a clear understanding :

1. Were the EFM configurations kept same in both the test cases i.e on Suse and RHEL environments ?

2. Was the chkconfig set for starting the database on both servers ? To be more specific, on Suse server was the database started once the server was up and running and on RHEL the database did not start ?

There might be a possibility of below scenario :

On RHEL (where the failover happened) after reboot of the server the database was not in running state, whereas on Suse (where failover did not happen) after reboot the database must be up and running.