anders.synstad

How to kill your nas ...

Discussion created by anders.synstad on Apr 3, 2014

Appears there is a stack buffer overrun error in the nas related to source
fields, that causes the nas to segfault. So, you can kill your nas with a
simple nimalarm command:

 

/opt/nimsoft/bin/nimalarm -S "$(for i in {1..550}; do echo -n d; done)" derp

 

Of course, I would strongly advise you not to run it.

 

This adds the long source to the NAS_NAME_SERVICE table in the sqlite
database.db that comes with the nas.

 

The nas kinda™ dies if source is > 511 chars, but REALLY dies if > 550 chars.


Should you happen to run into this problem, and see that your nas is chain
restarting with no obvious error message, it can be resolved by using a sqlite
browser tool (i.e. http://sourceforge.net/projects/sqlitebrowser/) to remove
the long entry from probes/service/nas/database.db, and then use Dr Nimbus to
remove the first message in the nas queue, as the nas dies after it has added
the entry to the sqlite database, but before it has ack'ed the message from the
queue.

 

Have fun knowing you can break your entire alarm flow from any robot in your
infrastructure with a simple nimalarm command. It makes me sleep good at night.

Outcomes