Note: This is a public test instance of Red Hat Bugzilla. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback at bugzilla.redhat.com.

Bug 539563

Summary:	yum can hang forever over IPv6
Product:	[Fedora] Fedora	Reporter:	Wolfgang Rupprecht <wolfgang.rupprecht>
Component:	yum	Assignee:	Seth Vidal <skvidal>
Status:	CLOSED DUPLICATE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	low
Version:	12	CC:	ffesti, james.antill, jason, kdudka, maxamillion, pmatilai, tim.lauridsen
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-12-22 19:29:10 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Wolfgang Rupprecht 2009-11-20 15:32:22 UTC

Description of problem:
yum over ipv6 hangs regularly and will stay hung for over 9 hours.  It needs a timeout badly. 

Version-Release number of selected component (if applicable):
$ yum --version
3.2.25
  Installed: rpm-4.7.1-6.fc12.x86_64 at 2009-10-28 21:18
  Built    : Fedora Project at 2009-09-21 13:30
  Committed: Panu Matilainen <pmatilai> at 2009-09-21 12:00

  Installed: yum-3.2.25-1.fc12.noarch at 2009-10-28 22:36
  Built    : Fedora Project at 2009-10-16 20:44
  Committed: Seth Vidal <skvidal at fedoraproject.org> at 2009-10-14 12:00

$ netstat -A inet6
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address               Foreign Address             State      
getnameinfo failed
tcp       38      0 sonic-tun.wsrcc.com:52973   [UNKNOWN]:https             CLOSE_WAIT  
getnameinfo failed
tcp        0      0 sonic-tun.wsrcc.com:40378   [UNKNOWN]:http              ESTABLISHED 
getnameinfo failed
tcp       38      0 sonic-tun.wsrcc.com:56893   [UNKNOWN]:https             CLOSE_WAIT  
tcp        0      0 sonic-tun.wsrcc.com:58909   mirror.steadfast.net:http   ESTABLISHED 

$ netstat -A inet6 -n
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address               Foreign Address             State      
tcp       38      0 2001:5a8:4:7d2::1:52973     2610:28:200:1::fed0:2:443   CLOSE_WAIT  
tcp        0      0 2001:5a8:4:7d2::1:40378     2001:18b8:1::beef:80        ESTABLISHED 
tcp       38      0 2001:5a8:4:7d2::1:56893     2610:28:200:1::fed0:2:443   CLOSE_WAIT  
tcp        0      0 2001:5a8:4:7d2::1:58909     2607:f128::3:80             ESTABLISHED 

Additionally, it looks like the mirrors are missing rDNS entries for their IPv6 addresses. 

How reproducible:
very.

Steps to Reproduce:
1.  yum list
2.
3.
  
Actual results:
infinite hangs if one uses the wrong mirrors

Expected results:
dead of buggy mirrors should time out.

Additional info:

Comment 1 seth vidal 2009-11-20 16:14:19 UTC

can you come by #fedora-admin on irc.freenode.org - some of the folks there who work on our ipv6 mirrorlist info would like to see if you can run a few tests to see where it is hanging up.

Comment 2 Wolfgang Rupprecht 2009-11-20 16:29:08 UTC

I'll try to connect up on irc a bit later.  In the mean time, this may give some folks an idea what is happening.

Bug #536775 (https://bugzilla.redhat.com/show_bug.cgi?id=536775) is a wireshark trace of a similar hang.  The first packet from the mirror after the 3-way handshake is lost.  Things come to a grinding halt.  This particular hang was for 9 hours before I forcefully stopped it by killing yum.

Comment 3 Jason Merrill 2009-12-22 15:20:16 UTC

I also find that ipv6 issues make yum very slow a lot of the time, though for me it usually only hangs for a few minutes sitting in SYN_SENT.

Comment 4 seth vidal 2009-12-22 19:29:10 UTC


*** This bug has been marked as a duplicate of bug 548269 ***

Comment 5 Wolfgang Rupprecht 2009-12-22 22:40:06 UTC

This is NOT really a duplicate of  548269.   As you can see from the trace, I have a valid and active IPv6 connection, yum simply hangs forever if the mirror stops sending.  Fixing this by needlessly disabling IPv6 would be a pretty heavy-handed fix.  All yum needs is a working timeout.

Of course someone could also look at the underlying IPv6 connectivity issue.  IPv6 works in general, but IPv6 to the Fedora mirrors seems problematic at best.

Comment 6 Kamil Dudka 2009-12-22 22:51:36 UTC


*** This bug has been marked as a duplicate of bug 520189 ***