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

Conference lassie::ucx

Title:DEC TCP/IP Services for OpenVMS
Notice:Note 2-SSB Kits, 3-FT Kits, 4-Patch Info, 7-QAR System
Moderator:ucxaxp.ucx.lkg.dec.com::TIBBERT
Created:Thu Nov 17 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:5568
Total number of notes:21492

5357.0. "UCX$LPD_queue processes 1 job per 3 minutes." by BACHUS::KNEUTS (Erwin Kneuts DTN 856-8726) Wed Mar 19 1997 13:59

Hello,

UCX 4.1 ECO 1 on vms 6.1:

I see that jobs in the UCX$LPD_queue remain PROCESSING for EXACTLY 
3 minutes 0 seconds.  The jobs are all 1 pagers.

The jobs are then sent to LAT terminal queues. Once the job is in the terminal
queue, it takes normal processing time (7 - 10 seconds) and it prints correctly.

The ucx$lpd_queue process takes a couple of CPU ticks when the job starts, and
then waits (LEF state) during 3 minutes.

This used to work fine until a couple of hours ago, and the customer did not 
reconfigure/change anything as usual :-)

I have been looking all over the place  for a 3 minute timer.  

A quick hint is highly appreciated as this is highly impacting (capacity = 20
jobs an hour, they need over 100) .  Up to now they have to keep up the pace
using manual requeue, but the guy's fingers start smoking.

Thanks,

Erwin
T.RTitleUserPersonal
Name		DateLines
5357.1It works again, but why...BACHUS::KNEUTSErwin Kneuts DTN 856-8726Wed Mar 19 1997 18:4711
- stop/start LPD : no change
- reboot cluster : no change 
- tried to replace ucx$lpd_smb and ucx$lpd_shr with 4.1 plain versions	
  failed for unclear reason. now nothing works at all in LPD.
- tired, fed up, wanna go home--> instal original 4.1 ECO 4 images again
- stop/start lpd: all works fine again.

I would appreciate to get some advice for breakfast tomorrow.


Erwin
5357.2DNS server was the culprit, but why do we need the DNS server AFTERWARDS?BACHUS::KNEUTSErwin Kneuts DTN 856-8726Fri Mar 21 1997 07:5029
Hi,

The problem came back, and we found the cause but do not understand
entirely the mechanism:

Here is the mechanism:

1) LPD client sends .cf and .df file to LPD server.   
    .cf file contains, among other info, the name of the client.
   LPD client has an entry for LPD server in host database.

2) LPD client (ucx$lpd_queue process) contacts DNS server, supposedly for
   backtranslation of incoming client IP address.
   When DNS server not available, time-out after 3 minutes, and 
   print job completes. During these 3 minutes the UCX$lpd_queue process
   has a BG device in the busy state.

If  the back translation is needed for security reasons, then I would expect 
no completion at all, otherwise it does not make sense.

If it is not for security reasons, I do not understand why the backtranslation
is needed, as the info is in the .cf file. 

The customer claims this is a UCX 4.1 behaviour, as he never experienced this
with 3.2, 3.3.  (and his DNS server has always been shaky)

Comments please?

Erwin
5357.3see discussion in note 374.*BACHUS::KNEUTSErwin Kneuts DTN 856-8726Mon Mar 24 1997 05:0910
The discussion in 374.* suggests this is  expected behaviour.

Personnally, I think the behaviour is inconsistent.

OR, the lpd request is rejected with a clear error message,
OR the lpd request is accepted and printed right away.

It does not make sense to print after 3 minutes  anyway.

Erwin