That's exactly the behavior I was trying to describe in my notes a couple weeks ago when Rob had a problem with FTP. For our customers, it was mostly a performance problem. IPV6 looks up "localhost" on off-the-box DNS servers *every time* if you don't have IPV6 fully configured including an IPV6 hosts file. That'll time out a lot of connections, or outright fail them if an IPV6 DNS server can't be found.
In short, if you can't take the time to fully configure IPV6, TURN IT OFF.
Apparently, with new PTF's, more and more services are being configured to "favor" IPV6 like java does. If IPV6 isn't running, the services will drop back to IPV4 and they work just like they always have.
-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of DrFranken
Sent: Tuesday, January 29, 2013 8:02 PM
To: Midrange Systems Technical Discussion
Subject: loopback or loopback which will connect?
Background: IBM i 7.1 partition with LOTS of PHP using the previously included Easycom DB connector. Because all database access is local every connection is defined to 'localhost'. This of course makes the code easily portable to other servers and means any changes to IP won't affect these connections. All the code is working marvelously and has been for some time on POWER5.
Three weeks ago we applied the current stack of PTFs to the partition.
After the IPL some scripts failed. Just some and in each case the failure was the connection to the database. Since nothing had changed this was very puzzling. Developers finally tried replacing loopback with
127.0.0.1 in the failing scripts. This worked. After fixing a few scripts, all seemed well.
A couple weeks later we took a full save and restored it to POWER7.
Nothing worked at all. Well that turned out to be the Easycom software needing a key, simple enough to fix that but even after that more scripts were failing with the connection problem but this time it was many more. We decided to fall back to the POWER5 server which was still running.
After bringing up the partition from dedicated system virtually EVERY DB connection was now failing. In every case replacing localhost with
127.0.0.1 solved the problem. Traces were sent to IBM who's primary response has been: "Everything looks correct."
After a few days we discovered a note about loopback and IPV6. This brought to mind some posts on Midrange-L that referenced incorrect configuration of IPV6 causing problems. So we simply shut down the IPV6 loopback since we weren't using IPV6 and had Never configured any of it.
Shazaam! Both the POWER7 and POWER5 servers immediately behaved correctly and all scrip references to localhost work. We've informed IBM but so far they've not responded to that piece of information.
Apparently some connections attempted to get to the IPV6 loopback and something in the PTFs and/or IPL caused more or fewer connections to attempt IPV6 vs IPV4 connections.
So two things:
1) Have others ran into this seemingly random connection failure problem with loopback while IPV6 is active?
2) A suggestion that if you are NOT using IPV6 turning it off to prevent that land-mine from being discovered.
- Larry "DrFranken" Bolhuis
www.frankeni.com
www.iDevCloud.com
www.iInTheCloud.com
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit:
http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at
http://archive.midrange.com/midrange-l.
midrange.com
