Topic: Something went wrong on the upgrade to 8.0 from 7.6 ...

[ricks1950]

The upgrade itself went perfectly smoothly, all packages installed.  The machine hangs on the final reconfigure, no disk activity, nothing.  After an hour of waiting, I booted the server, and it came up with no WAN -- configured it, and the machine again hung on the reconfig.  <CTRL><ALT><DEL> it boots, works, but warns that it needs a reconfigure, and has updates that need to be applied.  Machine is working, we have Internet, mail appears to be working, but any little thing will push this over the edge.

How do I get out of this?  I have a current (upgrade) make that backup on a USB stick, so reinstalling from scratch is not out of the question.

[CharlieBrady]

Please report any and all problems via the bug tracker. Thanks.

[ricks1950]

I have created the bug, and my system is still in its unstable state. 

What do I need to do to get the system back to its usual, reliable, stable self?

[CharlieBrady]

What do I need to do to get the system back to its usual, reliable, stable self?

1. Diagnose what is wrong.
2. Fix what is wrong.

[CharlieBrady]

The upgrade itself went perfectly smoothly, all packages installed.  The machine hangs on the final reconfigure, no disk activity, nothing.  After an hour of waiting, I booted the server...

That was the wrong thing to do. At that point, it would have been useful to type Alt-F2 and try to diagnose what was wrong. Perhaps 'top' would show you.

As a precaution, I would recommend you to memory test this machine.

[CharlieBrady]

... but any little thing will push this over the edge.

Perhaps you can better describe what you mean here.

[ricks1950]

Ran a memory test, passed OK, booted, hung on the "activating configuration" blue screen again, <alt>F2 log in, run top, shows there is a zombie process, but cpu usage is at 0%, top is the program using the most.  Hunted around for a while, did not see it.  Out of communication with the machine down, so brought it up to write this. 

If you want to give me some clues what I am looking for, I would be very happy to look for it.  Remember, I do not write software for a living, don't spend a lot of time fixing or installing SME server, more familiar with Debian and Ubuntu. 

I am not fully positive on what is fully configured, not confident on security or stability, and if it resets while I am not here to nurse it, it will not come up on its own.  It seems to work on the second reboot, and warns that it needs to be reconfigured, but hangs when you do.  Not quite the second circle of Hell, but, you get the picture.

While
"1. Diagnose what is wrong.
2. Fix what is wrong."
is great advice, it isn't very helpful.  It is what I want to do, with the advice of someone who knows this system a little more intimately than I do.

[ricks1950]

Ok, I have been picking through the problems in the server, and have gotten the final reconfigure to run.  The machine was trying to run a missing executable from an uninstalled package (Mailman) -- after several hours of trial and error, fixed it. Now I am trying to run an update, and am getting this circular set of errors. 

dmraid-events-1.0.0.rc13-65.el5.i386 from installed has depsolving problems
  --> Missing Dependency: dmraid = 1.0.0.rc13-65.el5 is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)
dmraid-events-1.0.0.rc13-65.el5.i386 from installed has depsolving problems
  --> Missing Dependency: libdmraid.so.1.0.0.rc13-17 is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)
dmraid-events-1.0.0.rc13-65.el5.i386 from installed has depsolving problems
  --> Missing Dependency: libdmraid.so.1.0.0.rc13-17(Base) is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)
Error: Missing Dependency: libdmraid.so.1.0.0.rc13-17 is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)
Error: Missing Dependency: libdmraid.so.1.0.0.rc13-17(Base) is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)
Error: Missing Dependency: dmraid = 1.0.0.rc13-65.el5 is needed by package dmraid-events-1.0.0.rc13-65.el5.i386 (installed)

I have exhausted my knowledge of Yum trying to fix dependencies.  Both packages are installed, and find locates libdmraid.so.1.0.0.rc13-17 -- Guidance please?
 

[ricks1950]

OK, so I worked another hour or three on this, and the system is now upgraded, updated, and running error free.  The dmraid package was being called from centos vault repository, removed that, and away we went.

Now, the system will reset whenever there is a USB drive or thumb drive connected to the PC.  Plug in the drive -- reset -- get to the initialize drives part of the reboot, and it resets continually.  Remove the USB drive, and the PC comes up normally.  This kind of blows up my back up scheme that I have used for several years. 

This part of USB has worked for so long, anyone have any ideas on how to approach fixing it?

[janet]

ricks1950

Try the process of elimination.
Remove the hard disks from that server, connect a clean empty hard disk and do a fresh install of sme8.0 from CD & do not install any contribs etc. Check that your USB works normally (ie correctly) or maybe it still plays up.

Doing this will check that ALL your hardware is compatible wirh sme8.0.
Have you searched the net for hardware compatibility ?

If everything is OK with the clean install then you have eliminated your harware as being the problem, so then you can start seriously troubleshooting the upgraded server.
More on that later.

Did you lodge a bug report as requested ?
SME server should upgrade without problems but it is wise to follow some basic procedures and concepts.
Did you check for and remove incompatible contribs etc ?

[ricks1950]

Firstly, this server has been in continuous service with SME on it since 2003, and it could not be more plain vanilla.  It had two contribs on it at one time or another, Mailman and rootkit hunter.  I removed both before the upgrade, mailman several months ago.  The first problem occurred because Mailman did not uninstall cleanly.  Second problem -- the sources did not get set properly, and had to be reset. 

The machine has been using USB drives for backup since version 7.0 came out.  At the point where the machine would not reconfigure, its backup ran, as scheduled.

There is a setting around the USB that has changed.

If I have to, I will find it myself, like every other step of this process, but I was hoping for some assistance.

[janet]

ricks1950

I was not joking when I gave you that advice, or assistance as you are calling it/requesting.

SME7.x to SME8.x has moved from Centos 4.x to Centos 5.x with many major changes.

Hardware & operating system compatibility is a basic requirement. On an old server like you have this could be a very significant factor. I was strongly suggesting to rule out compatibility issues before doing further troubleshooting as doing so could save a lot of time from being wasted. You could even have faulty hardware on such an old machine, and the power supply would be a likely/possible problem that could cause a shutdown or reboot when additional load is placed on the power supply voltage busses.

I suggest you look in the log files eg messages log file initially for any clues, but as the machine is resetting in an uncontrolled manner, then it may be hard to find pertinent info in any of the log files.

If you are not prepared to take my advice then I am wasting my time helping you.

Your conclusions (re setting change for USB) are guesswork as are mine at the present moment, without further testing investigation & analysis etc.

[ricks1950]

A couple of things went wrong with this upgrade: A contrib, Mailman, uninstalled several months ago, did not uninstall cleanly.  After the upgrade, this prevented the reconfigure from completing.  I found the point it was hanging in a log, and found the non-existant file being called in a script and removed the reference.  The machine just sat there waiting for something to happen that was never going to happen ... I wish I had written all this down to give you the details, but I didn't; had a brief look through the old logs looking for the needle in the haystack, if I have time later I will find it and update you.

Apparently, when I <cntrl><alt><del> to get out of the blue screen of the reconfigure, I broke something.  The sources for SME 7.6 as well as 8.0 were all enabled, so when the reconfigure finally ran, updates were available, but the updates would not complete, as there are conflicting packages between the releases.

I fixed this, ran the update to its successful conclusion, but am left with the USB issue.  I'm guessing that there is conflicting package or kernel module causing the issue, as this did not exist before the reconfigure successfully ran. 

This morning, in my admin email, I got this:

===
=== yum reports available updates:
===
Obsoleting Packages
giflib.i386                            4.1.3-7.3.3.el5                 base     
    libungif.i386                      4.1.3-1.el4.2                   installed
giflib-utils.i386                      4.1.3-7.3.3.el5                 base     
    libungif-progs.i386                4.1.3-1.el4.2                   installed
jwhois.i386                            3.2.3-12.el5                    base     
    whois.i386                         1.0.10-1                        installed
kernel-headers.i386                    2.6.18-308.8.2.el5              updates 
    glibc-kernheaders.i386             2.4-9.1.103.EL                  installed

My best guess is that the ultimate fix is to do a complete back up and reinstall and restore. 

To answer your concerns, my power supply is six months old, 450 watt name brand.  Machine is clean and dry in an air conditioned space (my home).  The USB is failing when the drive attempts to mount.  Other USB devices do not crash the system, as they do not attempt to mount.  The back up to USB ran the night the maching ran without the final reconfigure, so clearly, SME 8 can support the drive, and all the hardware.  I created a bug report. 

I have never thought about the reconfigure process before.  You see a blue screen in the middle of a boot sequence, no progress messages, I never knew (or cared) one way or another what state the machine was in when it was doing its thing.  Usually, I run headless, and this is done from a web browser or ssh shell. 

Those of us who do not live, eat and breathe SME server are not as aware of these things as Mary and Charlie are.  I thought I asked fairly clear questions, and got quite cryptic answers. 

[ricks1950]

Just to follow up, I just finished a fresh install of SME 8.0, and restored my backup data, and the system appears completely healthy (he says hopefully ... )

Just a couple of burps with replacing looser passwords inherited from years passed, but overall, should have done this before wasting 20 hours or so troubleshooting a failed upgrade.  All the email appears to be intact -- that was my biggest concern.

I know, I know, upgrades should work, and do most of the time, but in this case it did not. 

Thanks to those who attempted to help.