I applied it to our test bed. The combo installer hung during “Optimizing Volume”. Had to quit the installer, reboot, and run the Combo installer again.
Things seem to be in order now, but it’s not under the same loads as our production environment.
I also ran the 10.3.9 Combo updater on my home/test server box. It too got stuck at “Optimising Volume”, it kept spinning the candy bar without estimating a percentile. The Log showed that everything else went just fine so I quit the installer rebooted and ran ONYX to do a perm_repair, update_prebinding and some miscellaneous cleaning stuff, rebooted again.
The server seems snappier (it’s running on a Cube 450 with 576 Mb ram).
• Squirrelmail is a _lot_ faster than before, even without using the ‘performance cache’ option. I’m not really sure why though…
• I’m not running the AFP548 timsieved/avelsieve combo on this box so I don’t know whether these are broken…
• Cyrus seems happy, I didn’t get any errors on my server nor on my client machine.
• NFS performance seems to be slightly better.
So far so good, I don’t feel comfortable enough to run this update on any of my production servers yet, especially after the previous 2005-003 v1 update ordeal…, but this seems promising.
No fsck or permisions check beforehand, I did that a few weeks ago after hasseling about for three hours after the last 10.3.8 2005-003 update…
Even if I had done it, that shouldn’t prevent a _hanging_ Apple supplied installer. My guess is that the post-flight script does something funny which screws something up for the update_prebinding fase. (that’s two ‘somethings’ in one guess d-) A clue might be that all my services stopped responding just before the update_prebinding started, ssh and nfs died and so did the usual suspects (mail, web, etc.).
the update killed my server. obliterated the web config and server admin bombed big time. combo update did nothing. “java -version” returned a “segmentation fault” instead of a version #… had to rebuild it, but it was my test server. it was running blojsom, from afp548 article.
I upgrade my server (DP G5) without any problem and ran the upgrade on a new Mac mini. And since this time my remote home users cannot log in from this machine. the local admin account work fine.
What happen on the server is that the remote users get a kerberos ticket:
——–
Apr 18 13:03:47 server krb5kdc[427]: AS_REQ (6 etypes {18 16 23 1 3 2}) 123.123.123.23: ISSUE: authtime 1113822227, etypes {rep=16 tkt=16 ses=16}, [email protected] for krbtgt/[email protected]
Apr 18 13:03:48 server krb5kdc[427]: TGS_REQ (5 etypes {16 23 1 3 2}) 123.123.123.23: ISSUE: authtime 1113822227, etypes {rep=16 tkt=16 ses=16}, [email protected] for afpserver/[email protected]
——–
But on the machine the users do not pass through the login screen.
The “working” bar run forever with the text below “Logging in….”
On the client the system.log indicate that automount can’ t mount the remote server users home disk “authentication error (80)”
but after kerberos start and AFP_VPS mount the remote disk !!!
Well I ran a repair permission, reset the PRAM… but it’s still the same….
The strange thing is that the mac mini wasn’t seeing the update through the software update utilities from apple. So I download the .dmg directly from the web site of apple.
For those of you that have some problems with jave, probably everybody does, you have to reinstall the SecurityUpdate “SecUpd2005-002Pan.dmg”. After doing so, java will work correctly again.
Comments are closed