MySQL Bugs: #26581: server update doesn't return control when finished, but system is up and running

Bug #26581	server update doesn't return control when finished, but system is up and running
Submitted:	22 Feb 2007 23:32	Modified:	12 Nov 2009 9:16
Reporter:	Carsten Segieth	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Enterprise Monitor: Installing	Severity:	S3 (Non-critical)
Version:	1.1.0.4810 - 1.1.0.4869	OS:	Linux (Linux)
Assigned to:	BitRock Merlin	CPU Architecture:	Any
Tags:	UPDATE

Description:
At the end of the server update (as root user), after the last message, the control was not returned for >> 2 min (then I stopped waiting by pressing CTRL_C). Before cancelling I tried with hitting .ENTER. (nothing happened, that's why there are some blank lines below) and I logged in and checked that the system is up and running.

----------------------------------------------------------------------------
Setup has finished upgrading MySQL Network Service Manager Update on your computer.

Restart MySQL Network Service Manager now [Y/n]:

/opt/mysql/network/monitoring/upd_100_110r4810/mysqlnetworkctl.sh : mysql  started
Starting mysqld daemon with databases from /opt/mysql/network/monitoring/upd_100_110r4810/mysql/data/
Using CATALINA_BASE:   /opt/mysql/network/monitoring/upd_100_110r4810/apache-tomcat
Using CATALINA_HOME:   /opt/mysql/network/monitoring/upd_100_110r4810/apache-tomcat
Using CATALINA_TMPDIR: /opt/mysql/network/monitoring/upd_100_110r4810/apache-tomcat/temp
Using JRE_HOME:       /opt/mysql/network/monitoring/upd_100_110r4810/java
Syntax OK
/opt/mysql/network/monitoring/upd_100_110r4810/mysqlnetworkctl.sh : httpd started

qa-srv-b:/users/csegieth/mysql/network/monitoring #

How to repeat:
- see above
- I was running this update test in a screen session

Suggested fix:
return control to user

just had the same situation with a server update as a simple user, on the same box

just verified as simple user on RH4-x86 (I used net-qa1):

/users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/mysqlnetworkctl.sh : mysql  started
Starting mysqld daemon with databases from /users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/mysql/data/
Using CATALINA_BASE:   /users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/apache-tomcat
Using CATALINA_HOME:   /users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/apache-tomcat
Using CATALINA_TMPDIR: /users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/apache-tomcat/temp
Using JRE_HOME:       /users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/java
Syntax OK
/users/csegieth/mysql/network/monitoring/net-qa1/fresh_110r4869/mysqlnetworkctl.sh : httpd started

... and when this 'httpd started' appeared I waited a few minutes, logged into dashboard, and after this I used CTRL_C to stop the hanging update.

We have done another full round of testing (with and without running 'screen', just in case) but we are unable to reproduce.  What happens when you use mysqlnetworkctl.sh to start or stop the server from the command line. Does it hang at all?   Any other particularity about your environment that may help us reproduce this issue?

Can you send us a list of your environment variables? Just wanting to see if there is anything in there that may give us a clue.

Just got it again with an update 1.0.0 to current 1.1.0.4861, on "net-qa1" (RH4_x86) as user "root" (with sudo su root). Here my current env:

_=/bin/env
BK_CONFIG=license: BKL..license.removed.here..==;logging: none
BK_HOME=mysql.com
BK_LICENSE=ACCEPTED
BK_USER=csegieth
G_BROKEN_FILENAMES=1
HISTSIZE=1000
HOME=/root
HOSTNAME=net-qa1.mysql.com
HOST=net-qa1
INPUTRC=/etc/inputrc
KDEDIR=/usr
LANG=en_US.UTF-8
LESSOPEN=|/usr/bin/lesspipe.sh %s
LOGNAME=root
LS_COLORS=
MAIL=/var/spool/mail/csegieth
OLDPWD=/users/csegieth/mysql/network/agent/1.0.0/rhas4-x86/net-qa1/etc
PATH=/usr/kerberos/sbin:/bin:/usr/kerberos/bin:/usr/local/bin:/bin:/usr/bin:/usr/X11R6/bin:~/bin
PS1=\u@\h:\w>
PWD=/opt/mysql/network/monitoring
SHELL=/bin/bash
SHLVL=2
SSH_CLIENT=::ffff:10.100.64.58 4737 22
SSH_CONNECTION=::ffff:10.100.64.58 4737 ::ffff:10.100.1.224 22
SSH_TTY=/dev/pts/0
SUDO_COMMAND=/bin/su root
SUDO_GID=8964
SUDO_UID=8964
SUDO_USER=csegieth
TERM=cygwin
USER=root

... and here the exact command I used to start the update:

/users/schilders/public_html/official_builds/*official/archive/1.1.0.4869/service-manager/mysqlnetwork-*1.1.0.4869*-linux-update-installer.bin --mode text --installdir /opt/mysql/network/monitoring/upd_100_110r4869

Perhaps as a first help add one line 'Update has finished.' to the end of the script. This would make it more clear that one can use CTRL_C now ...

We have tried installing from a cygwin console but again we've been unable to reproduce it. Could please provide us with the exact binaries you are using for both server and update?

Deferring to 1.1.1