| Title: | DCE Product Information |
| Notice: | Kit Info - See 2.*-4.* |
| Moderator: | TUXEDO::MAZZAFERRO |
| Created: | Fri Jun 26 1992 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2269 |
| Total number of notes: | 10003 |
My customer is experiencing more delays with DCE. The system is a
2100A running DU v4.0b with C2 and DCE v2.0a with SIA. When the
customer does a "finger <username>" it takes 41 seconds to respond!
During that period 5200 packets were sent between our 2100A and
Security Server on an AIX box. 1700+ were sent from the 2100A and
3500+ were sent from the Security Server. So far, logging in takes
over 30 seconds, su takes about the same, and now 41 seconds for a
finger call.
Would someone please explain to me WHY this delay is there? Is it a
configuration problem at the customer or is this yet another v2.0a bug.
Please advise. If we (Digital) do not resolve this, we stand to lose a
lot of business.
If you need more information, please let me know.
Thank you.
Scott Fafrak
[email protected]
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 2266.1 | TUXEDO::CHUBB | Fri May 30 1997 10:30 | 6 | ||
Is DCE/SIA turned on? This would affect fingering time, as a lookup on
the user must happen through DCE. If SIA is off, then DCE should have
no affect. Also, how many users does your customer have in the
registry?
-- brandon
| |||||
| 2266.2 | SIA is on. Delay is too long. | DAGWST::FAFRAK | RISCy business... | Fri May 30 1997 13:54 | 16 |
DCE/SIA is definitely turned on (it is a customer requirement). There
several hundred accounts in the registry (soon to be thousands). I
think that the customer expects a little delay, but 30-40+ seconds is
NOT acceptable.
Would you please explain what it is doing that takes so long? BTW, a
straight finger w/o arguments returns immediately. But one with a
username as an argument incurs the delay.
Why so much traffic to and from the security server? This sounds to me
like something VERY inefficient is going on.
Thanks.
Scott
[email protected]
| |||||
| 2266.3 | /etc/sia/matrix.conf | VIRGIN::BILL | BILL is my lastname !!! | Tue Jun 03 1997 11:30 | 7 |
Can you please state the file pointed by /etc/sia/matrix.conf, just to make sure that this file isn't screwed up! Marco | |||||
| 2266.4 | matrix.conf is... | DAGWST::FAFRAK | RISCy business... | Tue Jun 03 1997 21:36 | 78 |
Marco,
The link is:
/etc/sia/matrix.conf -> /etc/sia/DCE_matrix.conf
I have included the matrix.conf file below. I really need to resolve
this problem soon. I have two VERY unhappy customers.
Should I QAR this?
Scott
[email protected]
Here is the file:
__________________begin included___________________________
#
# *****************************************************************
# * *
# * Copyright (c) Digital Equipment Corporation, 1991, 1996 *
# * *
# * All Rights Reserved. Unpublished rights reserved under *
# * the copyright laws of the United States. *
# * *
# * The software contained on this media is proprietary to *
# * and embodies the confidential technology of Digital *
# * Equipment Corporation. Possession, use, duplication or *
# * dissemination of the software and media is authorized only *
# * pursuant to a valid written license from Digital Equipment *
# * Corporation. *
# * *
# * RESTRICTED RIGHTS LEGEND Use, duplication, or disclosure *
# * by the U.S. Government is subject to restrictions as set *
# * forth in Subparagraph (c)(1)(ii) of DFARS 252.227-7013, *
# * or in FAR 52.227-19, as applicable. *
# * *
# *****************************************************************
#
# HISTORY
#
# @(#)$RCSfile: OSFC2_matrix.conf,v $ $Revision: 1.1.9.2 $ (DEC) $Date:
1996/02/
09 19:08:40 $
#
siad_init=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_chk_invoker=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so
)
siad_ses_init=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_ses_authent=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so
)
siad_ses_estab=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_ses_launch=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_ses_suauthent=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.
so)
siad_ses_reauthent=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.
so)
siad_chg_finger=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_chg_password=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.s
o)
siad_chg_shell=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_getpwent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_getpwuid=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_getpwnam=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_setpwent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_endpwent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_getgrent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_getgrgid=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_getgrnam=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_setgrent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_endgrent=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
siad_ses_release=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so
)
siad_chk_user=(DCE,/usr/shlib/libdcesiad.so),(OSFC2,/usr/shlib/libsecurity.so)
siad_get_groups=(DCE,/usr/shlib/libdcesiad.so),(BSD,libc.so)
| |||||
| 2266.5 | DCE SIA can be slow | TUXEDO::HASBROUCK | Wed Jun 04 1997 09:16 | 12 | |
This sounds to me like typical DCE SIA performance. The Digital UNIX DCE Product Guide has a section on SIA performance, if you haven't seen it. There was a major performance improvement going between DCE V1.3 and DCE V2.0, because there's a new, optional, sia routine called siad_get_groups. Does the matrix.conf file on the customer's system have an entry for it? (You'll probably only see this if DCE SIA is enabled.) Brian | |||||
| 2266.6 | DCE SIA is too slow... | DAGWST::FAFRAK | RISCy business... | Wed Jun 04 1997 17:52 | 10 |
Yes, there is siad_get_groups in the matrix.conf file. I posted the
file in .4. If someone would let me have access to the source,
maybe I can come up with a workaround for my customer. 30-42 seconds
is really too slow to be of much use. It is not that slow on our
competitor's implementations.
Thanks.
Scott
[email protected]
| |||||