Upgrade from 3.1 STD --> 3.2 Enterprise (?)

[jrmarino]

I've been running Litespeed 3.x for the last several months during our offseason to evaluate its performance. The standard version performed very well and I just ordered a permanent license for the enterprise version.

I have looked at the install/upgrade documentation on this website, but it's not immediately obvious to me what I need to do to upgrade. I want to transition to occur as quickly as possible as litespeed 3.1 standard is being used on our production server.

Is there a relatively seamless way to migrate/upgrade from standard to enterprise, or do I need to completely install 3.2 enterprise from scratch and copy over all the configuration, etc? FYI, I'm using Solaris 10 on X64_86 Sun servers.

 

[pguy]

Run the install script and it will give you the option of an install and an upgrade. Select upgrade and proceed. That should work. I too upgraded from the trial version to the paid ver this way

 

[jrmarino]

Okay, I tried a slightly different tactic.
I downloaded the 3.2 enterprise version for solaris X64_86 and exploded it.
I added the serial.no file that I created with the serial number I received in the mail.
I then executed install.sh

Then I found problems -- here's the transcript:

Found old configuration file under destination directory /opt/lsws.

To upgrade, press 'Enter', current configuration will not be changed.
To reinstall, press 'R' or 'r'.
To change directory, press 'C' or 'c'.

Would you like to Upgrade, Reinstall or Change directory [U/r/c]? U

Serial number is available.
Would you like to register a license key for this server? [Y/n]Y

Contacting licensing server ...

ld.so.1: lshttpd: fatal: libstdc++.so.6: open failed: No such file or directory
Killed

[OK] License key recived.
ld.so.1: lshttpd: fatal: libstdc++.so.6: open failed: No such file or directory
Killed
The license key recevied does not work.
Serial number is available.
Would you like to generate a license key request? [Y/n]Y

ld.so.1: lshttpd: fatal: libstdc++.so.6: open failed: No such file or directory
Killed

Did I make a mistake?
What should I do now?

 

[jrmarino]

By the way, that little trick knocked my working 3.2 standard version offline and it was down several minutes until somebody reported it.

Kind of exactly the opposite type of migration I was trying to achieve.

 

[jrmarino]

The solaris SMF log says litespeed dumped core (twice) so it turned off the process. I could not find the new core dump under /tmp/lshttpd/
I just restarted litespeed and it seems to be going again.

 

[jrmarino]

this is where libstdc++.so.6 is located on my server:

/usr/sfw/lib/amd64/libstdc++.so.6
/usr/sfw/lib/libstdc++.so.6

How do I make litespeed know where this library is?
This library appears to be needed for something specific to the licensing scheme as litespeed 3.x standard works on this server. Can the final license be generated manually to avoid this library?

 

[mistwang]

If you copy /usr/sfw/lib/amd64/libstc++.so.6 over to /lib/64/, it should work.

But, anyway we rebuild the 3.2 package with stdc++ library statically linked, so the problem should be gone. Please download the package again.

 

[jrmarino]

Thanks, the new build worked.

You should know this though: I answered "do not restart" litespeed when the install script asked (very last question) but was surprised to see litespeed 3.2 enterprise running immediately when I checked the server. It dumped core again and the watchdog restarted the server as 3.2 enterprise. See log:

SMF activity log (partial):

[ Jul 29 13:24:27 Method "start" exited with status 0 ]
[ Jul 30 10:11:07 Stopping because process dumped core. ]
[ Jul 30 10:11:07 Executing stop method ("/opt/lsws/bin/lswsctrl stop") ]
[OK] lshttpd: stopped.
[ Jul 30 10:11:08 Method "stop" exited with status 0 ]
[ Jul 30 10:11:08 Executing start method ("/opt/lsws/bin/lswsctrl start") ]
[OK] lshttpd: pid=27869.