Forums » Enduro/X other questions »

Problem during getting_started experiment on Ubuntu 20.04

Added by aka over 2 years ago

Hi,

I want to evaluate capability of Enduro/X vs Tuxedo by myself.

At first contact i tried to implement the getting_started sample manual.

But cant achieve to start anything :-\

I use the latest binary release for UBUNTU (despite my box is a focal fosa 20.04)
First step achieve well but whent trying "xadmin start -y" i have an error message

xadmin Log file seems point me to a problem with system limits adjustement that i have changed according to admin manual even if it's unclear for my to know what is mandatory or not (at a point it's says jump to next chapter in the doc if ubuntu, but does it mean next paragraph ?)

So, log shows complaining about Too many files open know and report queue config problem in the terminal output.

Could somebody gave me some guidance and let me start my discovery journey :-)

Best regards

user1@mypcbox:/opt/app1/log$ xadmin start -y
Enduro/X 7.5.22, build Mar 30 2021 20:32:54, using epoll for LINUX (64 bits)

Enduro/X Middleware Platform for Distributed Transaction Processing
Copyright (C) 2009-2016 ATR Baltic Ltd.
Copyright (C) 2017-2021 Mavimax Ltd. All Rights Reserved.

This software is released under one of the following licenses:
AGPLv3 (with Java and Go exceptions) or Mavimax license for commercial use.

ERROR ! Queue error, check logs

xadmin.log Magnifier (7.66 KB)


Replies (3)

RE: Problem during getting_started experiment on Ubuntu 20.04 - Added by Madars over 2 years ago

Hello,

Thanks for a question.

Usually for new OS you need to do following (for Linux) from the https://www.endurox.org/dokuwiki/doku.php?id=endurox:v7.5.x:guides:ex_adminman

1. Increase file limits:

Increase OS limits
$ sudo -s
# cat << EOF >> /etc/security/limits.conf

# Do not limit message Q Count.
# Some Linux 3.x series kernels have a bug, that limits 1024
# queues for one system user.
# In 2.6.x and 4.x this is fixed, to have
# unlimited count of queues (memory limit).
# ealrier and later Linuxes have fixed this issue.
*               soft    msgqueue        -1
*               hard    msgqueue        -1

# Increase the number of open files
*               soft    nofile  1024
*               hard    nofile  65536

EOF
# exit
$

2. Increase queue limits:

Setting Posix queue limits
Next step is to configure queue limits, this is done by changing Linux kernel parameters, in persistent way, so that new settings are applied at the OS boot.

$ sudo -s

# cat << EOF >> /etc/sysctl.conf

# Max Messages in Queue
fs.mqueue.msg_max=10000

# Max message size, to pass unit tests, use 1M+1K
fs.mqueue.msgsize_max=1049600

# Max number of queues system-wide
fs.mqueue.queues_max=10000

EOF

# Apply kernel parameters now
$ sudo sysctl -f /etc/sysctl.conf

# to check the values, use (print all) and use grep to find:
$ sudo sysctl -a | grep msgsize_max

Maybe try to reboot if not sure if these have applied.

If still have an issue, try to validate:

$ sudo sysctl -a | grep msg_max
$ sudo sysctl -a | grep msgsize_max
$ sudo sysctl -a | grep queues_max

that theses matches the desired settings.

Hope this helps.

RE: Problem during getting_started experiment on Ubuntu 20.04 - Added by aka over 2 years ago

Thank you Madars !

I was missing msgqueue entries in /etc/security/limits.conf and my adjustments for nofile were different, (too low) so like you suggest it works yeah :-)

user1@mypcbox:/opt/app1/conf$ xadmin start -y
Enduro/X 7.5.22, build Mar 30 2021 20:32:54, using epoll for LINUX (64 bits)

Enduro/X Middleware Platform for Distributed Transaction Processing
Copyright (C) 2009-2016 ATR Baltic Ltd.
Copyright (C) 2017-2021 Mavimax Ltd. All Rights Reserved.

This software is released under one of the following licenses:
AGPLv3 (with Java and Go exceptions) or Mavimax license for commercial use.

* Shared resources opened...
* Enduro/X back-end (ndrxd) is not running
* ndrxd PID (from PID file): 2399
* ndrxd idle instance started.
exec tpadmsv -k 0myWI5nu -i 10 -e /opt/app1/log/tpadmsv.log -r --  :
    process id=2402 ... Started.
exec tpadmsv -k 0myWI5nu -i 11 -e /opt/app1/log/tpadmsv.log -r --  :
    process id=2403 ... Started.
exec banksv -k 0myWI5nu -i 1 -e /opt/app1/log/BANKSV -r --  :
    process id=2404 ... Started.
exec banksv -k 0myWI5nu -i 2 -e /opt/app1/log/BANKSV -r --  :
    process id=2405 ... Started.
Startup finished. 4 processes started.

I was so exciting to test your work that i certainly read the manual in lazy mode.

Thanks again for this quick answer.

Best regards
ioda

(1-3/3)