|
Description
|
[Moved description where it belongs.]
I BFUed a test machine with on10 nightly archives. After rebooting, the
machine wasn't responding to pings.
SunOS Release 5.10 Version gate:2004-11-17 32-bit
Copyright 1983-2004 xxxxx Inc. All rights reserved.
Use is subject to license terms.
DEBUG enabled
Hostname: magneto
Configuring devices.
WARNING: asy1: could not map UART registers @ 0
WARNING: asy1: could not map UART registers @ 0
NIS domain name is it.sfbay.sun.com
Loading smf(5) service descriptions: 1/12
t_optmgmt: System error: Cannot assign requested address
12/12
Nov 18 11:59:53 svc.startd[100004]: Transitioning svc:/network/ntp:default
to maintenance due to misconfiguration.
magneto console login:
Nov 18 12:02:33 svc.startd[100004]: svc:/network/inetd-upgrade:default:
Method or service exit timed out. Killing contract 43.
Nov 18 12:02:33 svc.startd[100004]: svc:/network/inetd-upgrade:default:
Method "/lib/svc/method/inetd-upgrade start" failed due to signal Killed.
Nov 18 12:02:33 svc.startd[100004]: network/inetd-upgrade:default failed
At least it gave me a console login. All sorts of services were
complaining, but there seemed to be two or three root causes:
- ntp's failure mode looks like bug 6178183.
- network/service:default's start method was still running, and was blocked
running "/usr/sbin/ifconfig -auD4 netmask + broadcast +". Its service
log showed everyone's favorite message, "WARNING: Timed out waiting for
NIS to come up". I don't know if that's due to the t_optmgmt error or
what; "ifconfig -a" suggests that the interface isn't even plumbed.
This is probably a separate bug, but I don't have enough data to say why
the interface failed to come up. Any suggestions?
- inetd-upgrade's start method timed out after two minutes, so none of the
inetd services started. I cleared inetd-upgrade out of maintenance mode,
and died the same way. I looked at it while it was going, and it seems
pretty clear why it's wedged:
# uname -a
SunOS magneto 5.10 gate:2004-11-17 i86pc i386 i86pc
# ptree `pgrep inetd-upgrade`
100004 /lib/svc/bin/svc.startd
100571 /usr/bin/sh /lib/svc/method/inetd-upgrade start
100608 /usr/sbin/inetconv
# pstack 100608
100608: /usr/sbin/inetconv
c671ecf5 nanosleep (803ea78, 803ea80)
c65d26fb __yp_dobind_cflookup (8066910, 803eb0c, 0) + 106
c65d40fc __yp_match_cflookup (8066910, c650476c, 80697c8, 6, ...) + df
c6503e8f _nss_nis_ypmatch (8066910, c650476c, 80697c8, 803eb6c, ...) + 32
c6503f2c _nss_nis_lookup (8066930, 803ec20, 1, c650476c, ...) + 31
c65035d8 getbyname (8066930, 803ec20) + 8d
c66c73a1 nss_search (c6601110, c659d224, 4, 803ec20) + 17d
c659d2b6 getrpcbyname_r (80697c8, 803ec7c, 803ec88, 40c) + 69
c6773684 get_rpc_prognum (80697c8) + 2f
c67728e7 valid_props (806c670, 0, 0, 0, 0) + 191
08051fc0 ???????? (8069708, 8067110)
080523af ???????? (8069708, 8067110)
080527af ???????? (8067110, 1)
08053d4b main (1, 8047e94, 8047e9c) + 1a1
08051a96 ???????? (1, 8047f1c, 0, 8047f2f, 8047f47, 8047f73)
It looks like it's trying to go over the wire to look up an RPC program
number. This seems like an unwise behavior, especially for something
this early in boot. My "rpc" entry in /etc/nsswitch.conf is the default
one, namely "nis [NOTFOUND=return] files".
I've left the machine in its current state. If anyone wants access to it,
please let me know.
s7u4->s10_74L2nightly (1/9) LU using dvdnet on a ultra 5 system.
I found the following error messages from /var/adm/messages file.
Jan 9 15:52:04 line1-u5 svc.startd[7]: [ID 122153 daemon.warning]
svc:/network/inetd-upgrade:default: Method or service exit timed out. Killing contract 35.
Jan 9 15:52:04 line1-u5 svc.startd[7]: [ID 636263 daemon.warning]
svc:/network/inetd-upgrade:default: Method "/lib/svc/method/inetd-upgrade start" failed due to signal KILL.
The system works fine. I didn't see any name service problem or other problems.
I had also seen the error messages on the following test case on the same system.
s7u4->s1074l1 upgrade with dsr using cdrom
I had done s7u4->s10_74l1 nightly, s10_74l1 and s10_74l2 nightly LU and upgrade on the system about seven times and I only saw the message twice.
xxxxx@xxxxx.com 2005-1-10 20:42:43 GMT
I saw the following error messages on the following test case on Ultra 10.
s8hw4->s1074L2 nightly(1/11) upgrade ttinstall with dsr using cd net image
Jan 11 17:46:22 line1-u10 svc.startd[7]: [ID 122153 daemon.warning] svc:/network/inetd-upgrade:default: Method or service exit timed out. Killing contract 48.
Jan 11 17:46:22 line1-u10 svc.startd[7]: [ID 636263 daemon.warning] svc:/network/inetd-upgrade:default: Method "/lib/svc/method/inetd-upgrade start" failed due to signal KILL.
xxxxx@xxxxx.com 2005-1-13 00:15:53 GMT
I saw the following error messages on the following test case on Sun Blade 100.
s9u7->s1074L2 upgrade ttinstall with dsr using cdrom
Jan 16 15:43:34 line1-sb100 svc.startd[7]: [ID 122153 daemon.warning] svc:/network/inetd-upgrade:default: Method or service exit timed out. Killing contract 48.
Jan 16 15:43:34 line1-sb100 svc.startd[7]: [ID 636263 daemon.warning] svc:/network/inetd-upgrade:default: Method "/lib/svc/method/inetd-upgrade start" failed due to signal KILL.
This problem was still seen in upgrade S8U7->s10u4_05a. For the work around,
I used files instead of nis.
|
