Insufficient memory to create a queue

I'm a poor kid. I can't afford to buy extra memory for my queue. Too bad. And too sad.

After the root access issue has been resolved in WebSphere MQ Explorer, I am able to make a queue, but there is a prompt showing the following details to me:

****************************************
* Command: /opt/mqm/bin/crtmqm  Q1
****************************************
WebSphere MQ queue manager created.
Directory '/var/mqm/qmgrs/Q1' created.
The queue manager is associated with installation 'Installation1'.
AMQ6024: Insufficient resources are available to complete a system request.
exitvalue = 36 

I wasn't sure whether this is a general greeting or welcome message after a queue has been created? But I have verified the queue was there in the path /var/mqm/qmgr. To further verify this message isn't friendly to me, I issue another command sudo strmqm Q1, I got the same result:

$ sudo strmqm Q1
The system resource RLIMIT_NOFILE is set at an unusually low level for WebSphere MQ.
WebSphere MQ queue manager 'Q1' starting.
The queue manager is associated with installation 'Installation1'.
AMQ6024: Insufficient resources are available to complete a system request.

Clearly, this isn't a good thing. According to experts advice, there should be an error log locate at the path: /var/mqm/errors/. When I open the file located in that path, I see something like below:

+-----------------------------------------------------------------------------+
|                                                                             |
| WebSphere MQ First Failure Symptom Report                                   |
| =========================================                                   |
...
...
...
| Comment1          :- Failed to get memory segment: shmget(0x00000000,       |
|   73834496) [rc=-1 errno=22] Invalid argument                               |
| Comment2          :- Invalid argument                                       |
| Comment3          :- Configure kernel (for example, shmmax) to allow a      |
|   shared memory segment of at least 73834496 bytes                          |
|                                                                             |
+-----------------------------------------------------------------------------+

This reminds me there are additional settings for WebSphere MQ on Linux systems needs to be configured. According to the guide, there is minimum configuration are required for WebSphere MQ:

    kernel.shmmni = 4096
    kernel.shmall = 2097152
    kernel.shmmax = 268435456
    kernel.sem = 500 256000 250 1024
    fs.file-max = 524288
    kernel.pid-max = 120000
    kernel.threads-max = 48000

Among the settings, only shmmax and sem are not up to par. Below is what I have at current system:

$ cat /proc/sys/kernel/shmmax
33554432
$ cat /proc/sys/kernel/sem
250 32000 32 128

The following steps are what I did to fix this issue:

1. Open the file /etc/sysctl.conf.
2. Append the require configuration for shmmax and sem to the end of the file.
3. Reload the configuration with the command sysctl -p.

Where is MQ_INSTALLATION_PATH?

Sometimes I do wonder where will be the installation path is? There is a shortcut for this, issue a command dspmqver, it shows me the detail information about WebSphere MQ. From there I can see InstPath showing where this tool is installed:

Name:        WebSphere MQ
Version:     8.0.0.2
Level:       p800-002-150303.DE
BuildType:   IKAP - (Production)
Platform:    WebSphere MQ for Linux (x86-64 platform)
Mode:        64-bit
O/S:         Linux 3.13.0-62-generic
InstName:    Installation1
InstDesc:    
Primary:     No
InstPath:    /opt/mqm
DataPath:    /var/mqm
MaxCmdLevel: 801
LicenseType: Developer

Or maybe I have overlooked, the documentation did mention where does the MQ_INSTALLATION_PATH is. Below is the text extract from the guide:

The location where WebSphere MQ is installed is known as the MQ_INSTALLATION_PATH. The default location for the WebSphere MQ product code is shown in the following table:

Platform	Installation Location
Linux, HP-UX, and Solaris	/opt/mqm
AIX®	/usr/mqm
Windows 32-bit	C:\Program Files\IBM\WebSphere MQ
Windows 64-bit	C:\Program Files (x86)\IBM\WebSphere MQ

On UNIX and Linux systems, working data is stored in /var/mqm, but you cannot change this location.

WebSphere MQ Explorer requires root access

WebSphere MQ Explorer is my favorite tool as its ease of use compared to command line. This tool is a separate installation apart from the MQv8 installation, the component name would have this file name pattern MQSeriesExplorer_<suffix>-8.0.0-2.x86_64.rpm. Unfortunately, it requires root user which has mqm group granted in order to create a queue. Otherwise a prompt showing following error stopping me to proceed further.

****************************************
* Command: /opt/mqm/bin/crtmqm  Q1
****************************************
Access not permitted. You are not authorized to perform this operation. (AMQ4036)

Since this tool is installed in Ubuntu 14.04, and it was launched in Unity Launcher. Unlike command prompt, sudo would be my best friend handling root access control. Unfortunately, this is a graphical environment, I would need graphical sudo to help me. Here is the clue, open this file: /usr/share/applications/IBM_WebSphere_MQ_Explorer-Installation1.desktop (this file requires root access as well), and modify the line:

...
Exec=/opt/mqm/bin/MQExplorer
...

with this:

...
Exec=gksudo -k -u root /opt/mqm/bin/MQExplorer
...

Permission denied on crtmqm

This is so not good. I'm not able to create a queue after the WebSphere MQ installation. Following error was seen when I issue the command:

$ crtmqm Q1
bash: /usr/bin/crtmqm: Permission denied
$ sudo crtmqm Q1
AMQ7077: You are not authorized to perform the requested operation.

It has been so struggling with me at first, but later I have conquered the fear. As mention from this reference guide, I got a strong sense that I have not assigned my user ID to mqm group. Below is the text extract from the reference guide:

In WebSphere MQ, user id "mqm" and any ID which is a part of "mqm" group are the WebSphere MQ administrative users. WebSphere MQ queue manager resources are protected by authenticating against this user. Since the queue manager processes use and modify these queue manager resources, the queue manager processes will require "mqm" authority to access the resources. Hence, WebSphere MQ queue manager support processes are designed to run with the effective user-id of "mqm".

Since crtmqm is referring to /usr/bin/crtmqm, and it requires root access to it, thus it is wise to grand mqm group to the root user account instead of my user account.

MQv8 installation hit user limit error

Sigh~ There is another error while the MQv8 installation is undergoing. This time, the error was logged into a file, part of content causing the installation fails were shown here:

...

Current User Limits (root)
  nofile       (-Hn)  4096 files                         IBM>=10240        FAIL
  nofile       (-Sn)  1024 files                         IBM>=10240        FAIL

This has reminded me somewhere in the documentation did mention about the limit thing. Thus in my case, just append following configuration into /etc/security/limits.conf. This file may require root access in order to edit it.

...

mqm             hard    nofile          10240
mqm             soft    nofile          10240
# End of file

This should be the last error I face during the installation.

Cannot open Packages database during MQ installation

Continuing from the journey of previous installation on WebSphere MQ. This time, IBM has offered WebSphere MQ free of charge, the link of the site is here. As of this writing, I'm getting the MQv8 for Linux. No more waiting! This is a damn real great news for a poor guy like ME!

While I was about to install MQ after the step ./crtmqpkg suffix, something "abnormal" was blocking my way.

kokhoe@KOKHOE:/var/tmp/mq_rpms/LKH/x86_64$ sudo rpm -ivh --force-debian MQSeriesRuntime_LKH-8.0.0-2.x86_64.rpm 
error: db5 error(-30969) from dbenv->open: BDB0091 DB_VERSION_MISMATCH: Database environment version mismatch
error: cannot open Packages index using db5 -  (-30969)
error: cannot open Packages database in /home/kokhoe/.rpmdb
error: db5 error(-30969) from dbenv->open: BDB0091 DB_VERSION_MISMATCH: Database environment version mismatch
error: cannot open Packages database in /home/kokhoe/.rpmdb

I think this error doesn't belong to MQ specific because it happened on another package installer too. The clue I found was there is a hidden folder, .rpmdb is having some conflict since previous installation. By just removing that folder will make the rpm program smiling.

mqlicense unable to locate Java Program

In the recent WebSphere MQ version (v7.0.1.3 as of this writing) installation on Ubuntu 13.04, I have some difficulty issue in accepting the WebSphere license. The output shows that the JAVA program went missing. Below is the message being prompt for the missing JAVA program.

...

ERROR: No useable java programs at ./lap/jre/jre/bin/java 

ERROR:  Installation will not succeed unless the license
        agreement can be accepted.

        If the error was caused by a display problem,
        read the license agreement file  (LA_xx, where
        xx represents your language ) in the 'licenses'
        directory, and then run the following command:

            mqlicense.sh -accept

        Only use this command if you accept the license
        agreement.

        For other errors, contact your IBM support centre.

I find it interesting why the mqlicense shell script is looking for the JAVA program in ./lap/jre/jre/bin. My curiosity is leading me to dig into mqlicense.sh to find out what is going on to the script, and I found this piece:

...

# Set JRE location
  JRE=${JRE:-$(find $PROGPATH/lap -type d -name bin)/java }
  if [ ! -x ${JRE?} ]; then
    echo "ERROR: No useable java programs at ${JRE}"
    errormsg
    exit 1
  fi

...

I was so surprise that WebSphere isn't install the JAVA program in default location (/bin or /usr/bin). Anyhow I just change the JRE variable to suit my needs. Now the screen will turn out a license agreement page for me to accept. Once accepted, following message will be seen:

...
...

Displaying license agreement on :0

Agreement accepted:  Proceed with install.

apt-get gone crazy due to my mistake

I was installing WebSphere MQ for my own research. Due to some technical mistake, I have cause some trouble on apt-get command not able to function. Below is the error I got:

...
...

chown: invalid user: ‘mqm:mqm’
chown: invalid user: ‘mqm:mqm’
chown: invalid user: ‘mqm:mqm’
dpkg: error processing mqseriesserver (--configure):
 subprocess installed post-installation script returned error exit status 1
Errors were encountered while processing:
 mqseriesserver

Not knowing why this could happened to me? Due to my mistake I guess? Then I spot apt-get autoremove would lend me a help. What this command actually do?

autoremove is used to remove packages that were automatically installed to satisfy dependencies for other packages and are now no longer needed.

I did try on that but no luck. While I'm almost hopeless, I see some light, it is pointing to apt-get remove --purge. What this command actually do?

--purge
Use purge instead of remove for anything that would be removed. An asterisk ("*") will be displayed next to packages which are scheduled to be purged. remove --purge is equivalent to the purge command. Configuration Item: APT::Get::Purge.

For my case, I'm going to purge any of the mqseriesserver thing, thus by launching following command:

sudo apt-get remove --purge mqseriesserver

I can see the rubbish is being purge from apt-get command:

...

Removing mqseriesserver ...
Purging configuration files for mqseriesserver ...
Processing triggers for libc-bin ...
ldconfig deferred processing now taking place

Now my apt-get is back to normal. Phew~