Bug 1024676

Summary: Warnings in PostgreSQL 9.2 log
Product: OpenShift Online Reporter: Jarek Zgoda <jarek.zgoda>
Component: ImageAssignee: Paul Morie <pmorie>
Status: CLOSED UPSTREAM QA Contact: libra bugs <libra-bugs>
Severity: medium Docs Contact:
Priority: medium    
Version: 2.xCC: agrimm, ccoleman, cdaley, dmcphers, jeff.mccormick, jhonce, simon.brandhof
Target Milestone: ---Keywords: SupportQuestion, UpcomingRelease
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2014-02-21 17:34:07 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Jarek Zgoda 2013-10-30 08:44:23 UTC 
Related forum post: https://www.openshift.com/forums/openshift/warnings-in-postgresql-log

Description of problem:
postgresql-9.2 cartridge in application brewlog-zgoda.rhcloud.com

After startup of cartridge there are following entries in log

LOG:  could not bind socket for statistics collector: Permission denied
LOG:  trying another address for the statistics collector
LOG:  could not bind socket for statistics collector: Permission denied
LOG:  trying another address for the statistics collector
LOG:  could not bind socket for statistics collector: Cannot assign requested address
LOG:  trying another address for the statistics collector
LOG:  could not bind socket for statistics collector: Cannot assign requested address
LOG:  disabling statistics collector for lack of working socket
WARNING:  autovacuum not started because of misconfiguration
HINT:  Enable the "track_counts" option.

PostgreSQL is running without autovacuum process:

[brewlog-zgoda.rhcloud.com log]\> ps -A f | grep postgres
 1932 ?        S      0:00 /opt/rh/postgresql92/root/usr/bin/postgres
 1939 ?        Ss     0:00  \_ postgres: logger process                  
 1941 ?        Ss     0:00  \_ postgres: checkpointer process            
 1942 ?        Ss     0:00  \_ postgres: writer process                  
 1943 ?        Ss     0:00  \_ postgres: wal writer process              
13879 ?        Ss     0:00  \_ postgres: username dbname 127.8.158.130(21744) idle
14161 ?        Ss     0:00  \_ postgres: username dbname 127.8.158.130(21775) idle

Version-Release number of selected component (if applicable):


How reproducible:
always in this application

Steps to Reproduce:
1. create application with MySQL cartridge and start it
2. add postgresql-9.2 cartridge
3. remove MySQL cartridge
4. restart application observing PostgreSQL log

Actual results:
PostgreSQL is running without autovacuum process

Expected results:
Autovacuum process is present

Additional info:
Tried and not working:
 - remove postgresql-9.2 cartridge then add it again
Comment 1 Jarek Zgoda 2013-10-30 15:51:36 UTC 
Exactly same result can be observed with completely fresh application, just python-2.7 and postgresql-9.2 cartridges.
Comment 4 Jeff McCormick 2014-01-23 21:27:59 UTC 
just wondering about status of this bug. The issue seems to be in how localhost is locked down.  Can the internal IP address that is assigned by openshift be added to the list of IP addresses for localhost within /etc/hosts?

according to my collegue, Greg Smith, "What PostgreSQL does at startup is ask the OS for the addresses of "localhost" (hard-coded) using the getaddrinfo() system call: https://en.wikipedia.org/wiki/Getaddrinfo

It then loops over all the returned addresses until it successfully talks to socket it created, and if all addresses fail it gives the error in the bug report."
Comment 5 Clayton Coleman 2014-01-24 21:22:55 UTC 
Vacuum is pretty important.  Can we get this fixed in postgres?
Comment 6 Paul Morie 2014-01-27 18:47:50 UTC 
I've created a bug for this w/ Postgres.  Postgresql uses a mailing list for bug reports instead of an issue tracker; my report is current pending moderator approval.  Will link once it has been posted.
Comment 7 Paul Morie 2014-02-21 17:34:07 UTC 
Unfortunately it doesn't look like this is something we'll be able to fix right now.  We hope that we'll be able to support it in the future.
Comment 8 Andy Grimm 2015-02-10 19:18:56 UTC 
Just to connect the dots, here's what Paul filed upstream:
http://www.postgresql.org/message-id/20140127183125.26701.88068@wrigleys.postgresql.org

Given Tom Lane's response, I don't think postgres is going to change behavior:

""""
Yeah.  In particular, I'd always supposed that a platform that wanted
to use some other IP address than 127.0.0.1 for loopback would also be
expected to make sure that "localhost" mapped to that other IP address.
Otherwise, there simply isn't any way to make network-using software
work except random hacks.

Frankly, I don't think you're going to get any cooperation from Postgres
in adapting to such a broken networking environment as this, and I doubt
other upstreams are going to be any more positive about it.
"""
Comment 9 Simon Brandhof 2015-02-21 12:41:57 UTC 
Paul, you mention in http://www.postgresql.org/message-id/1241903739.14805210.1390938412164.JavaMail.root@redhat.com that using network namespaces in OpenShift is a high priority improvement. Where can its status be tracked ? Is there a bugzilla bug ?
Thank you