[Yum] Strange timeout issue...

[Cris Rhea]

I'm running a CentOS 6/x86_64 system with all updates.

The yum config points at a local CentOS repo mirror:

    [base]
    name=CentOS-$releasever - Base
    #mirrorlist=http://mirrorlist.centos.org/?release=$releasever&arch=$basearch&repo=os
    baseurl=ftp://clustersw-212.mayo.edu/CentOS/$releasever/os/$basearch/
    gpgcheck=1
    gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-6

(same idea for "updates", "extras", and "contrib" repos)

Running:

# yum update
Failed to set locale, defaulting to C
Loaded plugins: fastestmirror, security
Loading mirror speeds from cached hostfile
ftp://clustersw-212.mayo.edu/CentOS/6/os/x86_64/repodata/repomd.xml: [Errno 12] Timeout on ftp://clustersw-212.mayo.edu/CentOS/6/os/x86_64/repodata/repomd.xml: (28, 'Connection time-out')
Trying other mirror.
ftp://clustersw-212.mayo.edu/CentOS/6/extras/x86_64/repodata/repomd.xml: [Errno 12] Timeout on ftp://clustersw-212.mayo.edu/CentOS/6/extras/x86_64/repodata/repomd.xml: (28, 'Connection time-out')
Trying other mirror.
ftp://clustersw-212.mayo.edu/CentOS/6/updates/x86_64/repodata/repomd.xml: [Errno 12] Timeout on ftp://clustersw-212.mayo.edu/CentOS/6/updates/x86_64/repodata/repomd.xml: (28, 'Connection time-out')
Trying other mirror.
Setting up Update Process
No Packages marked for Update


But, wget or curl succeeds:

# wget ftp://clustersw-212.mayo.edu/CentOS/6/os/x86_64/repodata/repomd.xml
--2012-08-16 15:39:18--  ftp://clustersw-212.mayo.edu/CentOS/6/os/x86_64/repodata/repomd.xml
           => `repomd.xml'
Resolving clustersw-212.mayo.edu... 129.176.212.87
Connecting to clustersw-212.mayo.edu|129.176.212.87|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done.    ==> PWD ... done.
==> TYPE I ... done.  ==> CWD (1) /CentOS/6/os/x86_64/repodata ... done.
==> SIZE repomd.xml ... 3824
==> PASV ... done.    ==> RETR repomd.xml ... done.
Length: 3824 (3.7K) (unauthoritative)

100%[======================================>] 3,824       --.-K/s   in 0.02s   

2012-08-16 15:39:18 (231 KB/s) - `repomd.xml' saved [3824]


If I do an strace:

gettimeofday({1345149621, 565646}, NULL) = 0
clock_gettime(CLOCK_MONOTONIC, {83862, 679396531}) = 0
socket(PF_INET, SOCK_STREAM, IPPROTO_TCP) = 8
fcntl(8, F_GETFL)                       = 0x2 (flags O_RDWR)
fcntl(8, F_SETFL, O_RDWR|O_NONBLOCK)    = 0
connect(8, {sa_family=AF_INET, sin_port=htons(40181), sin_addr=inet_addr("129.176.212.87")}, 16) = -1 EINPROGRESS (Operation now in progress)
clock_gettime(CLOCK_MONOTONIC, {83862, 679916402}) = 0
poll([{fd=8, events=POLLOUT|POLLWRNORM}], 1, 29975

(hangs there until timeout)

(129.176.212.87 is the correct address for clustersw-212.mayo.edu)

The repo server works fine for all other clients using it and there's no 
issue network-wise getting from this machine to the repo server.

With all the timeouts, the "yum update" takes approx 1.5 mins to complete.

I suspect this is some sort of issue with libresolv and/or all the patches
RHEL has come out with recently regarding turning off IPV6 kernel modules 
(we're not using IPV6, nor round-robin on DNS servers).

Has anybody seen these symptoms before and if so, what's the cause/cure?

Thanks--

--- Cris

-- 
 Cristopher J. Rhea
 Mayo Clinic - Research Computing Facility
 200 First St SW, Rochester, MN 55905
 crhea at Mayo.EDU
 (507) 284-0587