[arm-allstar] Repeater key down lockup question

[Tom Whiteside]

We had a lockup occur during an ARES net last that has me scratching my
head.   We are running two arm-allstar repeaters here with a permanent link
and using version 1.0 of the software.   One has been up for almost 8 months
and the other for two months with an additional 5 months in test mode.
These are actively used repeaters logging typically 30 hours of transmit
time a month and they have been very solid.

So what I am about to report may be a fluke???   Last night's lockup was the
second one we have experienced - both have been on the machine that also
hosts Echolink and both at times with multiple Echolink users logged on.   I
point that out because mostly we no Echolink connections and then typically
only one while there were 5 logged in last night when the lockup occurred...
Not sure if that has anything to do with it but seems like an interesting
coincidence at least.

When the lock up occurred, a station was keyed up and talking and then the
audio disappeared - the station talking could still be heard on reverse.
When that station unkeyed, we did not hear any courtesy beep or sound of any
kind and the repeater remained keyed.   

I was able to log into the controller and initiated an astres.sh script but
nothing happened.   Fortunately, I have a remote outlet controller and
cycling power on the controller cleared the condition and the repeater
booted normally.

Several questions as a result of this:  

(1) I have been just letting these systems run without any periodic restarts
or reboots of the RP2 controllers.   The last restart of Asterisk was 31
days before this happened and no telling how long the RP2 had been up.
Should I be periodically restarting things for good measure?

(2) I'm not a LINUX experienced user - mostly just using what I have learned
on this project?   In Windows, I know where to look at event logs and task
manager to learn things about the state of things.   Is there an equivalent
set of things I should be check on the PI?

(3) Have you seen this sort of thing happen?   

(4) Do you think there is a correlation with added loading from Echolink?
Should I consider deploying an additional RP on a separate Allstar node to
handle the Echolink connections?

Tom N5TW