[Postgres-xl-general] pgxc_ctl : add slave datanode failed

Nikhil Sontakke nikkhils at gmail.com
Wed Apr 20 03:11:44 PDT 2016


Florian,

You mentioned the below:

"Bedides the log file, the on screen log shows that it can't ping (don't
know which host), but it's normal since postgres user has no this
permission on my system."

I think this is the reason for the pg_hba.conf corruption since as I
mentioned in my earlier email, the function internally uses "ping" to get
the IP address if the provided input happens to be a hostname. In your
case, the ping is failing returning some junk causing the pg_hba.conf
corruption.

Regards,
Nikhils

On Wed, Apr 20, 2016 at 3:36 PM, Nikhil Sontakke <nikkhils at gmail.com> wrote:

>
>
>> It is visible. Actually, if i understand how adding a slave works, the
>> master is shut down to update it's pg_hba.conf, then restarted and then,
>> pg_basebackup is called.
>>
>>
> Ah yes, indeed. That's the case. I realized that after writing the email.
>
>
>> However, since the modification of the master pg_hba.conf corrupts it,
>> the master can't be restarted, hence this error message. And indeed, before
>> adding the slave, the master is up and running, and after this error
>> message, the master is not running and its logs indicates the bad
>> configuration in pg_hba.conf
>>
>> Hmm, that looks like a bug then in the pgxc_ctl code where it's
> "constructing" the entry to add to pg_hba.conf of the master.
>
> <!-- digs around the code a bit.. >
>
> Seems like the getIpAddress() function is returning a spurious value on
> your setup causing the corruption in the master pg_hba.conf file.
>
> # Additional entry by adding the slave, 20160419_20:43:34
> host replication postgres 172.16.3.0/24 trust
> host replication postgres PJ�^A/32 trust
> # End of addition ===============================
>
>
> Can you please run this command from the host where you are running
> pgxc_ctl and check if the output is sane? This command is being used in the
> above function.
>
> ping -c1 172.16.3.15  | head -n 1 | sed 's/^[^(]*(\\([^)]*\\).*$/\\1/'
> Regards,
> Nikhils
>
>
>> I've tried to trace why the pg_hba gets corrupted, but didn't succeed yet
>>
>> thanks
>>
>> Florian
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.postgres-xl.org/pipermail/postgres-xl-general-postgres-xl.org/attachments/20160420/7ee289ac/attachment.htm>


More information about the postgres-xl-general mailing list