On 9/28/18 4:08 AM, Neal Becker wrote:
Matthew Miller wrote:
> You're not using spot pricing, are you?
>
No.
I'm becoming more convinced that machines get shutdown when load average >
#cores for too long, but it seems rather randomly. The reason I'm asking
here is just in case that is part of fedora cloud, rather than a function of
AWS.
And that's why I asked if you had looked at the AWS admin portal to see
if AWS is shutting down the instances and if so, why. Then again, they
MAY not tell you why they shut you down. That's happened to us and we're
still pretty miffed about it.
Assuming AWS isn't shutting you down, then you need to look at your
workloads and things. I suggested having a small instance with EBS
persistent storage to act as a log server and having your instance(s)
log to it so you can see if there's something in your workload that
would cause the instance(s) to self-destruct (like an OOM or something).
I sorta doubt there's anything inherent with Fedora Cloud itself that
would cause this, but I suppose there are corner cases that might
aggravate a condition that causes self destruct. You won't know unless
you eliminate AWS' terminations and/or log your activity to see what's
up.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- I'm afraid my karma just ran over your dogma -
----------------------------------------------------------------------