Thread: problems with postgresql 10.1 hba_conf on fedora 27

problems with postgresql 10.1 hba_conf on fedora 27

From
support-tiger
Date:
can someone please help ?

postgresql-10

fedora 27

our hba_conf that worked with pg 9x

local    all   all       postgres                peer

host    all    all       127.0.0.1/32         md5

host    all    all       ::1/128                  md5

now gives error on restart and we cannot find the  log file

postgres remains so damn difficult and time wasting to quickly get up 
and running  vs other db's and docs don't help much




-- 
Support Dept
Tiger Nassau, Inc.
www.tigernassau.com
406-624-9310




Re: problems with postgresql 10.1 hba_conf on fedora 27

From
"David G. Johnston"
Date:
On Wed, Dec 20, 2017 at 4:05 PM, support-tiger <support@tigernassau.com> wrote:
can someone please help ?

postgresql-10

fedora 27

​Installed via yum?
our hba_conf that worked with pg 9x

​There are six of them, which one specifically?

now gives error on restart and we cannot find the  log fil
​e

​And what error would that be?​

David J.

Re: problems with postgresql 10.1 hba_conf on fedora 27

From
James Keener
Date:
> postgres remains so damn difficult and time wasting to quickly get up
and running  vs other db's and docs don't help much

See, I feel exactly the opposite. The docs are some of the best of any database or open source software I've used. Postgres also does the "right" thing almost all the time, making it predictable.

> now gives error on restart and we cannot find the  log file

What error? Where did you look for the log files? Did you try running the deamon directly to see if that gives any additional information?

Jim

On December 20, 2017 6:05:39 PM EST, support-tiger <support@tigernassau.com> wrote:
can someone please help ?

postgresql-10

fedora 27

our hba_conf that worked with pg 9x

local    all   all       postgres                peer

host    all    all       127.0.0.1/32         md5

host    all    all       ::1/128                  md5

now gives error on restart and we cannot find the  log file

postgres remains so damn difficult and time wasting to quickly get up
and running  vs other db's and docs don't help much




--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: problems with postgresql 10.1 hba_conf on fedora 27

From
rob stone
Date:

On Wed, 2017-12-20 at 16:05 -0700, support-tiger wrote:
> can someone please help ?
> 
> postgresql-10
> 
> fedora 27
> 
> our hba_conf that worked with pg 9x
> 
> local    all   all       postgres                peer
> 
> host    all    all       127.0.0.1/32         md5
> 
> host    all    all       ::1/128                  md5
> 
> now gives error on restart and we cannot find the  log file
> 
> postgres remains so damn difficult and time wasting to quickly get
> up 
> and running  vs other db's and docs don't help much
> 
> 
> 
> 

Hi,

"local" is for socket connections, yet you have an address of
"postgres". How did this function in the past?

Exactly what do you mean by "gives error on restart". The database
starts and you can connect, then you shut it down and then you start it
again, it fails?

Cheers,
Rob


Re: problems with postgresql 10.1 hba_conf on fedora 27

From
"David G. Johnston"
Date:
On Wednesday, December 20, 2017, rob stone <floriparob@gmail.com> wrote:

> local    all   all       postgres                peer
>
> host    all    all       127.0.0.1/32         md5
>
> host    all    all       ::1/128                  md5
>
> now gives error on restart and we cannot find the  log file
>

"local" is for socket connections, yet you have an address of
"postgres". How did this function in the past?


Sorta, but correct enough.  Local lines and host lines have differing numbers of columns.  As written the auth-method is "Postgres" with auth-option of "md5" (i.e., there is no address column for local) which is simply wrong.  Whether it's an email typo or the actual file...

I suspect it's possible we used to ignore malformed lines whereas now we error...

David J.

Re: problems with postgresql 10.1 hba_conf on fedora 27

From
support-tiger
Date:



On 12/20/2017 06:33 PM, David G. Johnston wrote:
On Wednesday, December 20, 2017, rob stone <floriparob@gmail.com> wrote:

> local    all   all       postgres                peer
>
> host    all    all       127.0.0.1/32         md5
>
> host    all    all       ::1/128                  md5
>
> now gives error on restart and we cannot find the  log file
>

"local" is for socket connections, yet you have an address of
"postgres". How did this function in the past?


Sorta, but correct enough.  Local lines and host lines have differing numbers of columns.  As written the auth-method is "Postgres" with auth-option of "md5" (i.e., there is no address column for local) which is simply wrong.  Whether it's an email typo or the actual file...

bingo!  Thks - should be "local all postgres  peer"  (how many times we looked at this and didn't see it  (:   But if this setup is not good, we would like to know what it should be - our case is for accessing from Ruby app using pg gem on same server (at least for now)

No, the docs for understanding  hba_conf are not good (yes we can read and are fairly smart)  - we made suggestions the last time for several case examples and were ignored - okay, simplicity of  pouchdb/couchdb  is getting our attention








I suspect it's possible we used to ignore malformed lines whereas now we error...

David J.


-- 
Support Dept
Tiger Nassau, Inc.
www.tigernassau.com
406-624-9310

Re: problems with postgresql 10.1 hba_conf on fedora 27

From
Matt Zagrabelny
Date:


On Wed, Dec 20, 2017 at 9:30 PM, support-tiger <support@tigernassau.com> wrote:


No, the docs for understanding  hba_conf are not good (yes we can read and are fairly smart)  - we made suggestions the last time for several case examples and were ignored - okay, simplicity of  pouchdb/couchdb  is getting our attention


Can you point to a link where the hba docs are suboptimal and suggest to the mailing list an improvement?

-m