Hello,

I'm pleased to announce the release of SlapOS Node version 0.35.

As I haven't advertised release of SlapOS Node version 0.34, I will include it in this blog post.

Now that SlapOS Node is quite stable (since 0.33), the 0.34 and 0.35 releases basically contains many small improvements and simplifications that you may love to use. It also contains edge cases support in case of something being corrupted in a partition.

Feel free to update your SlapOS Node package in your favorite distribution.

Note: in Debian based distributions, upgrading the package may cause instances to change their IP. If you care about your current IPs, just backup the /opt/slapos/slapos.xml file into your home directory before upgrade, and place it back after upgrade.

Here is the changelog:

0.35 (2012-02-08)

-----------------

 * slapformat now creates a backup of old slapos.xml containing all IPs in case of future problem.

 * slapformat hopefully doesn't change local IPv4 of partition if IPv6 changed. 

 * SlapOS Node creates and runs a /opt/slapgrid/$MD5/buildout.cfg for each Software Release, for easier hacking/debugging.

 * SlapOS Node no longer crashes if sending an error to the SlapOS Master triggers itself an error.

 * SlapOS Node no longer crashes if partition's certificate does not exist on master.

 * Add a CLI query command for slapproxy (slapproxy-query).

 * Many minor other typo / output fixes.

0.34 (2013-01-23)

-----------------

 * All the logs are rotated and will be deleted after a few days. No more 500MB logs!

 * Binary cache now properly works with Ubuntu. Ubuntu is no longer recognised as "Debian/unstable".

 * SlapOS Client now correctly detects the default configuration file. It means running "slapos request X Y" works.

 * SlapOS Node no longer crashes if one partition is in bad state when reporting.

 * SlapOS Node no longer tries to force-process a stopped instance.

 * The Watchdog subsystem now correctly reports an error with the correct certificate.

 * SlapOS Node creates a symlink to software_release in each partition to ease debugging.

 * The "slapproxy" system now handle complex cases like resilient Software Releases and Slave instances.

 * Many minor other improvements, see "CHANGES.txt" of slapos.core.git for details.

Happy slapping,

The SlapOS Node Team

Let me introduce you the brand new watchdog.

Watchdog is a simple SlapOS Node feature allowing you to watch any process

managed by supervisor. Whenever this process exits, it will trigger an alert

(bang) that is sent to the SlapOS Master.

As a reminder, the "bang" cause tree recalculation. In human language, it means

"forcing buildout run on all instances of the service".

To watch a process, you will need two things:

1/ A recent SlapOS Node (>= 0.32.2)

2/ Populate the ${buildout:directory}/etc/service directory. Of course, scripts

meant to be exited (like erp5-bootstrap) still go to the ${buildout:directory}/etc/run directory.

Here is the related documentation: How to write Instance Profile

Happy watching!

Hello guys,

It's been a long time since I haven't advertised a new version of SlapOS Node (a.k.a slapgrid, in old naming).

In the last weeks, we published a lot of small incremental releases of slapos.core, always
more polished.

Today we released a new version of SlapOS Node. The whole changelog is here: http://git.erp5.org/gitweb/slapos.core.git/blob/d988134e9189bae9c042a16febff90862870af29:/CHANGES.txt.
(You can know the current version of slapos.core by spotting the latest version of slapos.core in /opt/slapos/eggs.)

Let me highlight the most important changes we made those last 2 months:

* Add a "watchdog" : if a watched process crash, it will trigger immediately a "bang". (I'll write more doc on this later)
* Added "periodicity" feature: if an instance has not been processed by slapgrid after defined time, process it.
* Introduce simpler "slapos" entry point. See here for documentation: http://git.erp5.org/gitweb/slapos.core.git/blob/HEAD:/documentation/slapos.usage.rst?js=1. It allows to do things like "slapos node software" to replace the looong older commands.

* slapgrid: Introduce --only_sr and --only_cp.
- only_sr filter and force the run of a single SR, and uses url_md5
(folder_id)
- only_cp filter which computer patition, will be runned. it can be a
list, splited by comma (slappartX,slappartY ...)
* Add "slave instances" support in slapproxy

* "-c" (--console) option is now useless: we ALWAYS output into stdout, and in file if --logfile is specified.
* Removed the random delay when running slapgrid
* Much less calls to master, and much more type checks compared to older versions. It means, no longer "computer_id=None" errors.

Most of those features are now daily tested.

Of course, in case of bug, you can slap me directly. Or use the forums.

To update, you can read the doc here : http://community.slapos.org/wiki/osoe-Lecture.SlapOS.Extended/slapos-howto-upgrade

In the very near future, we will continue to improve usability of the different slapos commands, and provide a nicer usage/documentation for it.

Happy upgrade!

SlapMonitor

Slapmonitor is a tiny python script located in slapos.toolbox. It is written to monitor CPU and RAM resources usage of Apache and MySQL applications and store this information  to existing logbox server like Wallix LogBox.

Documentation how to setup and configure it can be found here.