Bug 90996

Summary: up2date doesn't attempt to connect to host
Product: [Retired] Red Hat Linux Reporter: Graham Harris <pro>
Component: up2dateAssignee: Adrian Likins <alikins>
Status: CLOSED WORKSFORME QA Contact: Fanny Augustin <fmoquete>
Severity: medium Docs Contact:
Priority: medium    
Version: 7.1CC: gafton, mihai.ibanescu, pro
Target Milestone: ---   
Target Release: ---   
Hardware: i686   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2004-08-20 20:45:12 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 Graham Harris 2003-05-16 08:46:20 UTC
Description of problem:
up2date (command line) fails with 
"Error communicating with server.  The message was:
host not found".

up2date-gnome fails silently- no GUI appears.

In neither case is any traffic sent to the default IP route (as seen by 
Ethereal monitoring eth0
'route' shows this default route 

default    zlan   0.0.0.0         UG    0      0        0 eth0

and there are no other routes which could lead to redhat.com

up2date-gnome has been working successfully for a long while, stopped 10 
days or so ago. Touch anything? Moi?

Version-Release number of selected component (if applicable):
2.8.39

How reproducible:
Every time.

Steps to Reproduce:
1. Start Ethereal, start capturing on eth0
2. run up2date or up2date-gnome
3. wait for finish (dwell a decent pause for up2date-gnome) and stop capture
    
Actual results:
no traffic seen.

Expected results:
some https traffic in both directions

Additional info:
I can telnet from the affected host to the RHN server specified using 
up2date --configure

telnet www.rhns.redhat.com 443
Trying 66.187.232.101...
Connected to www.rhns.redhat.com.
Escape character is '^]'.

My ISP reputedly does no filtering in or out

Comment 1 Graham Harris 2003-05-16 08:51:06 UTC
rhnsd is running

RHN says that "System not checking in with RHN" since 2003-05-06 20:31:37 (EST)



Comment 2 Mark J. Cox 2003-05-16 10:21:45 UTC
check hostname used to contact rhn in /etc/sysconfig/rhn/up2date

Comment 3 Adrian Likins 2003-05-16 18:44:39 UTC
Versions in 7.1 should be attempting to reach

www.rhns.redhat.com

The error seemed to indicate a DNS error. Verify that
you can ping www.rhns.redhat.com

Comment 4 Graham Harris 2003-05-18 09:47:40 UTC
/etc/sysconfig/rhn/up2date contains




serverURL=https://www.rhns.redhat.com/XMLRPC




As above, I can 'telnet www.rhns.redhat.com 443'. I cannot 'ping www.rhns.
redhat.com' which I put down to that host refusing to respond to a ping, since 
telnet works. Anyway it can't be a DNS problem if telnet works. 




up2date still fails the same way




Thanks for your interest




Comment 5 Graham Harris 2003-05-30 18:11:39 UTC
I can also open the serverURL https://www.rhns.redhat.com/XMLRPC in my web 
browser. I get a "Method not allowed" page. Ethereal logs the transaction in 17 
packets (well, frames, to be pedantic).




Just to reiterate then: I can contact the server by many means from the same 
machine. up2date on the other hand does not contact the server- Ethereal cannot 
find any evidence that up2date is making any attempt to contact the server.




Hope you can help,


Thanks


Comment 6 Graham Harris 2003-06-12 20:32:00 UTC
I made ethereal look for traffic on any interface, not just the one that ought 
to forward traffic to the internet. "up2date" at the command line seems to 
generate a DNS query for "pro.spicedreams.com" which is the name of my system. 
It's a real global DNS name, for which my system is itself authoritative. So it 
looks like up2date sends a DNS query from 217.169.25.210 to 217.169.25.210 
asking for the IP address of pro.spicedreams.com and gets the response back from 
217.169.25.210 that the answer is "CNAME spicedreams.com A 217.169.25.210".




This is repeated verbatim about 28 seconds later (first time I tried- next time 
this step was omitted). 




Almost immediately, 127.0.0.1 tries to open a connection to port 512 on 127.0.0.
1 and gets "destination unreachable" in response. Might be unconnected with the 
foregoing- is this something you would expect from up2date, if so do I need to 
look at firewall rules again?




"update agent" in gnome doesn't appear even to generate this much traffic. 




Any suggestions as to what I can try next? 




I'm anxious the system doesn't get too far behind the pack. Is there a way I can 
diagnose the problem further, or update without up2date?




Comment 7 Graham Harris 2003-07-27 19:12:10 UTC
Guys, could you give me some idea how to progress this please?



Comment 8 Graham Harris 2003-10-06 19:31:50 UTC
OK, using strace the problem seems to be that a bunch of python source files are 
not found: 

readlink("/usr/bin/python", 0xbfffe030, 1024) = -1 EINVAL (Invalid argument)

exists, is the default python executable

stat64("/usr/bin/Modules/Setup", 0xbfffde80) = -1 ENOENT (No such file or direct
ory)

There is a /usr/lib/python1.5/config/Setup but no /usr/bin/Modules/Setup

stat64("/usr/bin/lib/python1.5/string.py", 0xbfffde60) = -1 ENOENT (No such file
 or directory)
stat64("/usr/bin/lib/python1.5/string.pyc", 0xbfffde60) = -1 ENOENT (No such fil
e or directory)

These two are actually in /usr/lib/python1.5 not /usr/bin/lib/python1.5

stat64("/usr/lib/python1.5/string.py", {st_mode=S_IFREG|0644, st_size=15430, ...
}) = 0
stat64("/usr/bin/Modules/Setup", 0xbfffde90) = -1 ENOENT (No such file or direct
ory)
stat64("/usr/bin/lib/python1.5/lib-dynload", 0xbfffde90) = -1 ENOENT (No such fi
le or directory)
stat64("/usr/lib/python1.5/lib-dynload", {st_mode=S_IFDIR|0755, st_size=4096, ..
.}) = 0
brk(0x80b5000)                          = 0x80b5000
stat64("/usr/lib/python1.5/exceptions", 0xbfffd9d0) = -1 ENOENT (No such file or
 directory)
open("/usr/lib/python1.5/exceptions.so", O_RDONLY) = -1 ENOENT (No such file or
directory)
open("/usr/lib/python1.5/exceptionsmodule.so", O_RDONLY) = -1 ENOENT (No such fi
le or directory)

Does this help?

Comment 9 Graham Harris 2003-10-09 11:42:21 UTC
Problem understood now.

It seems up2date REQUIRES nscd, which wasn't installed. Nothing else seems to 
require nscd, so everything apart from up2date continued working when nscd got 
removed from the system (it was not obvious that it was useful since gnoRPM says 
just that it "improves performance"). 

And when nscd is missing, up2date fails silently (gnome) or with an unhelpful 
error (cli).

Comment 10 Adrian Likins 2003-10-22 01:35:26 UTC
up2date nor anything else should require "nscd". I dont run
it on any of the systems I use.

Generally if something works with nscd, but fails without it,
it indicates DNS/name resolution confiration issues of some
sort. nscd just caches DNS and name services queries locally,
so I have no idea what the correlation would be. 

Comment 11 Graham Harris 2003-10-26 20:48:24 UTC
Thanks.

nscd seems to be the determinant. If nscd really is only caching then up2date 
has a real dependance upon it. If up2date has no dependance on nscd as you 
suggest, then nscd is masking an underlying dns config problem, ie nscd is not 
transparently caching.

How do we progress this? Redhat seems to blame the nscd maintainers- a tough 
call, given that you claim nscd is not necessary yet if nscd is absent, up2date 
fails. Do you want to raise it with the nscd folks?