[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference decwet::advfs_support

Title:AdvFS Support/Info/Questions Notefile
Notice:note 187 is Freq Asked Questions;note 7 is support policy
Moderator:DECWET::DADDAMIO
Created:Wed Jun 02 1993
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1077
Total number of notes:4417

1001.0. "dtadvfs error" by NETRIX::"johnson@alf.dec.com" (decatl::johnson) Wed Feb 12 1997 16:45

I am attempting to help in a dtadvfs issue and see that note 939.0 
has similar error:
(is there a dtadvfs notes conference?)

------------------------------------------------------------------
Wed Feb 12 14:51:14 1997 | AUDIT | WARNING - failed security check 
	| File socket_client.c | Line 374
Wed Feb 12 14:51:14 1997 | ERROR | Failed to connect to agent for host: 
	gila.tmp.com
	 | File app_agent.c | Line 183
Wed Feb 12 14:51:14 1997 | ERROR | Failed to make agent connection to host: 
	gila.tmp.com
	 | File agentConnectMainWindow.d | Line 241
----------------------------------------------------------------------


# hostname 
returns "gila.tmp.com" and host.allow file has this same name.

I cant seem to find anything named "agentConnectMainWindow" or "app_agent.c"
here in our Source pool at CSC/AT kernel team, so dont have any idea
where this is coming from. 

the customer (DEC) has talked to the networks folks, and us.
 
Based on the note and another call, I suggested he has a problem in his
network setup.

/etc/services has the proper  3 lines for advfs

fverify has 0 errors on the AFAADVGUI401 and AFAADVDAEMON401 subsets.

Any futher ideas or help would be appreciated.

sid johnson
Customer Support Center/Atlanta
UNIX KERNEL TEAM (and ADVFS)

[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name
DateLines
1001.1DECWET::MARTINThu Feb 13 1997 14:0215
What does the advfsd log file show?  All that dtadvfs can know is that it failed
the security check, advfsd should tell you why the security check failed.

What does gethostbyaddr() tell you about your IP address?  Does it return
"gila.tmp.com"?

You don't have the dtadvfs or advfsd source code at CSC/AT, as it is not (nor
can it be) checked in with the Digital UNIX ODE pools.

If they are not suffering any other network problems, then you should file a
QAR/CLD/IPMT against advfs_gui, and include the last few lines of both the
advfsd and the dtadvfs log files, and probably the contents of the
/var/opt/advfsd/socket/hosts.allow as well.

--Ken
1001.2thanksNETRIX::"johnson@alf.dec.com"decatl::johnsonFri Feb 14 1997 09:4521
thanks for reply, will follow suggestions. if not solved by network folks
(checking the domain lookups).

We found that 

	# nslookup    gila
		indicates   gila.tmp.com

	  ** but the servier is <server>.TMP.COM

server is SUN, so he is going back to networks to see is uppercase
is significant in resolution.  Or, if they should of used TMP.COM
on initial bind setup (yek)...

	tryed adding a alias  'gila.TMP.COM'

and didnt seem to help, but maybe the network folks knows another 
hint or 2.

sid johnson
[Posted by WWW Notes gateway]
1001.3DECWET::MARTINFri Feb 14 1997 14:1713
I don't know if it would be helpful or not, but advfsd compares the incoming
client address from accept(2), which it then uses

	inet_ntoa(3),
	gethostbyaddr(3)->h_name, and
	gethostbyaddr(3)->h_aliases

to compare these possibilities to all the entries in the hosts.allow file.

I've seen instances where hostname(1) reports something that is different from
anything that the two functions above return.

--Ken
1001.4Possible display issues that haven't been checkedLEXSS1::MACOMBERUNIX Software ConsultantTue Feb 18 1997 13:0612
Sid, et al,

 I happened to read note 991 talking about dtadvfs not appearing which got
me thinking about my scenario. It's possible, and I have calls into my peers
to double check this, that the machine in question may have had it's prior
life as an NT machine. Which, if logica follows, may mean that I may have a
display problem as well. I know that the graphics adapter on the machine was
changed prior to installing UNIX 4.0a, but I can't honestly say that ECU was
run afterwards "or" that the machine used to be an NT box. I shall check this
out.

/robb.gui.challenged
1001.5In addition to .-1LEXSS1::MACOMBERUNIX Software ConsultantTue Feb 18 1997 13:161
A couple more things:
1001.6Some more inputASABET::nqsrv113.nqo.dec.com::WorkbenchTue Feb 18 1997 15:2415
One of my collegues gave me the following input:

>That 1000 at Monster board was a NT system before I loaded unix. When I did 
>the install, the gui never came up due to a graphics compatability issue with 
>the inbedded graphics adapter. Don installed a PCI graphics card and got X to 
>start. 
>The next time I saw the system is when I walked in with you. The 1000 has the 
>firmware from the 3.7 or 3.8 CD and the ECU is rev 1.10, which is current.
>
>The system itself is a very old 1000, so there maybe some issues with the rev 
>of the motherboard.

For what it's worth.

/robb.still.notfixed
1001.7Presto! (wha happened?)ASABET::nqsrv217.nqo.dec.com::WorkbenchWed Feb 19 1997 22:1423
Well,

 A solution has been found. The problem with the solution is I can't figure 
out what is was that I did that fixed the problem. (drat!)

 Here's what I "thought" I did. When I visited the customer today, the first 
thing I did was look, once again, at the gui log file. Then I did an nslookup 
to verify that, in fact, gila.tmp.com was returned. Then I pinged both 
gila.tmp.com and gila's IP address, both sucessfully. I did and arp -a and saw 
appropriate entries. I then looked at the /etc/hosts file and, noticing a few 
oddities - like two entries for localhost and two entries for gila: one saying 
just gila and the other additionally specifying gila.tmp.com - I decided to 
review the entire hosts file. Just for clarity's sake I sorted the hosts file 
to get a resonable look at the numerical IP addressing. After closing the 
hosts file, I ran dtadvfs& again and guess what: it worked. I then put the 
original hosts file back in place and ran dtadvfs& again and it worked. I have 
no idea what changed.

 I was able to manipulate the /var/opt/advfsd/socket/hosts.allow file to 
manage the other host in the environment so I knew that hosts.allow was O.K. 
but I still have no idea what solved the problem. Ghost in the machine!?

/robb.glad.curious