Install Recommended Path Cluster Fails
When I try to install the patch cluster on a new install I get two kinds of failures depending on what I am trying to do.
I am installing on a 2900 and the Solaris 2.10 (3/05)
If I use the nosave option then I get a failure code of 45
If I don't use the nosave option I get this:
ld.so.1: zonename: fatal: relocation error: file /sbin/zonename: symbol zonecfg_is_scratch: referenced symbol not found
Can someone tell me what is going on?
[476 byte] By [
Arador68] at [2007-11-26 6:28:59]
# 1
I would recommend opening a ticket with Sun to investigate this issue.
# 2
Did you get an anwer on this? I'm running into something similar.
When I first ran 'install_cluster -nosave -q' every patch install was
erroring with status 45. After 10 or so failures in a row, I killed the
instal_cluster command. Now both the original command
and 'install_cluster' give the same zonename failure error you reported:
# uname -a
SunOS unknown 5.10 Generic sun4u sparc SUNW,Sun-Blade-100
unknown# ./install_cluster -nosave -q
Patch cluster install script for Solaris 10 Recommended Patch Cluster
The nosave option was used. Objects will not be saved.
ld.so.1: zonename: fatal: relocation error: file /sbin/zonename: symbol zonecfg_is_scratch: referenced symbol not found
It appears to be a shared object resolution problem. The zonename command
is indirectly trying to access libzonecfg.so.1, but that library is in /usr/lib:
# zonename
ld.so.1: zonename: fatal: relocation error: file /sbin/zonename: symbol zonecfg_is_scratch: referenced symbol not found
Killed
# truss /sbin/zonename
....
stat("./libzonecfg.so.1", 0xFFBFEA70)Err#2 ENOENT
ld.so.1: zonename: fatal: relocation error: file /sbin/zonename: symbol zonecfg_is_scratch: referenced symbol not foundwrite(2, " l d . s o . 1 :z o n".., 119)= 119
# find /usr/lib -name libzonecfg.so.1 -ls
19828 128 -rwxr-xr-x1 rootbin120800 Jan 22 2005 /usr/lib/sparcv9/libzonecfg.so.1
19827 112 -rwxr-xr-x1 rootbin104420 Jan 22 2005 /usr/lib/libzonecfg.so.1
# 3
As with the previous poster we would recommend that you open a ticket with Sun for this issue.
# 4
Hi,I am curious as to the status of this problem since I also the same issue? ThanksRich
# 5
Hi Rich,
I did a little more research and it looks like the problem is related to errors
installing a patch for libzonecfg (120900) or zoneadm(d) / zonecfg (121333).
Possible resolutions I've seen are:
1) Back out one or both patches mentioned above. I can't try this because I used
the nosave option when I tried to install the cluster.
2) Re-install the SUNWzoneu package. The problem has knocked my system off
the network so I'll have to try to find an OS CD to try this.....
Good luck,
John
# 6
FYI, here's the hack that enabled my system to boot multi-user
to get the system back on the network so I could obtain a copy
of the SUNWzoneu package for re-install. This worked because
I was not using zones on this system.
1) boot single user
2) '/' was mounted read-only, make it writable: mount -o remount /
3) create a new zonename command:
mv /usr/bin/zonename /usr/bin/zonename-
echo 'echo global' > /usr/bin/zonename
chmod 755 /usr/bin/zonename
4) reboot
# 7
I also faced same problem. Solution give by jhgage is working.Thanks Jhgage. Reason is /sbin/zonename is not reprting zonename. Now it is. But it is temp solution.When you use this command from Zone , you might be failed.Mahendra , India
# 8
I fell into the same blankety blank trap. *Several* lessons learned.
Some of the failing SMF scripts cited /sbin/zonename. I used your same scheme but at /sbin. Note, every one, it is [apostrophe]echo global[apostrophe], not [grave]. My browser font baaaaarely distinguishes them.
Kudos: I was very happy to have your guidance.
George Wyche
# 9
The workaround mentioned above worked for me.
Replacing /sbin/zonename with a script that echoes "global" got my system back up.
Then re-installing SUNWzoneu from the original media was the permanent fix that got everything working again and allowed me to put the real /sbin/zonename back in place.
odex at 2007-7-6 14:24:06 >
# 10
I had this problem too upgrading a Blade 1500 from 3/05 to 11/06, and these messages are the only reference to the problem I found.
The workaround I used was to simply install the SUNWzoneu package in single user mode, which also worked.
I'm guessing that if you know about this problem beforehand, at the end of the upgrade you could just pkgadd the SUNWzoneu package using the console that is left open. I will get the chance to try this rather soon......
