My 'minimum' instances keep restarting, is this normal?

[ChrisR]

My yaml file is as follows for one of my applications:

path: /hal/admin/rails/XmlTransfer
type: mongrel_rails
min_instances: 3
max_instances: 6
timeout: 600
user: root
group: root

My problem is that every 10 minutes the 3 minimum instances that are
running will automatically restart.

I am guessing that this is not expected behaviour, because If i were
to set my timeout to 9 seconds (as in the example yml files) then you
wouldn't expect all 3 min instances to restart every 9 seconds.

Any ideas what is wrong? Is anything wrong?

Heres my config.yml for good measure :

host: v20zserver
port: 10000
backend_port_start: 60000
backend_port_end: 61000
status_update: 5
inactivity_timeout: 800

Its not really a serious issue, and maybe its normal, but it makes me
feel wierd when I see all my applications continuously restarting
themselves every 5-10 minutes overnight.

Thanks
Chris

[Peter Cooper]

On Apr 2, 9:53 am, ChrisR <EvilGeen...@gmail.com> wrote:
> Its not really a serious issue, and maybe its normal, but it makes me
> feel wierd when I see all my applications continuously restarting
> themselves every 5-10 minutes overnight.

I need to get back into the code again as it's been a month since I've
worked on SwitchPipe, but I believe there is a secondary timeout for
instances like this. If you have, say, 1 minimum instance, that
instance should not run forever and it will be restarted from time to
time. I need to establish again exactly how that "time to time" is
defined :) It may well simply be ten minutes and the time is a
coincidence with your setup :)

Pete

[Peter Cooper]

Just had a look at the code, and any backend process that's an hour
old will be restarted (at least, as of the latest build). I just need
to check, but I assume this is not what you're seeing when you say "5
- 10 minutes"?

Short of a bug, which I will certainly try to find, the only other
thing I can think is that something is clogging up that first
"minimum" instance, and it's getting cleaned up by the timeout. That
would cause what you are seeing, and might be connected to the hanging
bug that was raised (and is now fixed) in another thread. Upgrade to
the latest trunk build and see if this error still occurs. If it does
not, that will be the issue.

Cheers,
Peter Cooper