You are viewing a plain text version of this content. The canonical link for it is here.
Posted to ruleqa@spamassassin.apache.org by Jari Fredriksson <ja...@iki.fi> on 2017/10/06 19:52:03 UTC

OOh: 96 cores on Google Cloud

I have had a VM in Google Cloud with 32 cores as a backup that my
raspberry automatically rises up and calls, if it can not reach my Core
i7 PC. And always on saturdays for net-jarif. 

Now Google announced that they have 96 cores available! 

I will now run 96 cores on weekdays (backup) and 192 threads for
net-jarif. This is amazing. 

And all orchestrated by a Raspberry Pi 2. 

-- 
jarif@iki.fi

Re: OOh: 96 cores on Google Cloud

Posted by Jari Fredriksson <ja...@iki.fi>.
Kevin A. McGrail kirjoitti 6.10.2017 22:57:
> On 10/6/2017 3:52 PM, Jari Fredriksson wrote:
>>=20
>> I have had a VM in Google Cloud with 32 cores as a backup that my=20
>> raspberry automatically rises up and calls, if it can not reach my=20
>> Core i7 PC. And always on saturdays for net-jarif.
>>=20
>> Now Google announced that they have 96 cores available!
>>=20
>> I will now run 96 cores on weekdays (backup) and 192 threads for=20
>> net-jarif. This is amazing.
>>=20
>> And all orchestrated by a Raspberry Pi 2.
>>=20
> That... Is.... Awesome...
>=20
> I just had a client that had a spin up systems on AWS with Skeddly to
> shutdown.=C2=A0 They missed the little checkbox on Skeddly that "this rul=
e
> is not a test".
>=20
> They hit a million boxes before AWS called them up and asked them if
> they were really trying to run a million boxes.=C2=A0 Don't be them :-)
>=20
> Regards,
> KAM

My script on the raspi has supported Azure (not used anymore by me) and=20
Cloud, and local core i7

It has these:

run_azure_masscheck()
{
     trap_add "azure vm shutdown sa-ruleqa" EXIT SIGHUP SIGINT SIGTERM
     azure vm start sa-ruleqa

and

connect_gcloud() {
     machine_type=3D$1
     if [ -z $machine_type ]; then
        echo "connect_gcloud machine_type"
        exit 2
     fi
     trap_add "gcloud/gcloud-sa-ruleqa.py delete" EXIT SIGHUP SIGINT=20
SIGTERM


if something goes berserk and scripts has a failure then the VM will be=20
usually shut down gracefully and I do not need to pay $4000 per month ;)

The cost for me has been so far between $1.5 - 2.0...

If someone is interest on a remote masschech I could share

00		17	* * Sat bin/auto-mass-check-remote.sh --gcloud-upload-corpus
00              19      * * *   test -x bin/delete-unwanted-mail.sh &&=20
bin/delete-unwanted-mail.sh
00              20-23/1 * * Mon,Tue,Wed,Thu,Fri,Sun     test -x=20
bin/auto-mass-check-remote.sh && flock -n .auto-mass-check-remote.lock=20
nice bin/auto-mass-check-remote.sh --auto-mass-check
00              20-23/1 * * Sat   test -x bin/auto-mass-check-remote.sh=20
&& flock -n .auto-mass-check-remote.lock nice=20
bin/auto-mass-check-remote.sh --gcloud-mass-check


br. jarif


--=20
jarif@iki.fi

Re: OOh: 96 cores on Google Cloud

Posted by "Kevin A. McGrail" <ke...@mcgrail.com>.
On 10/6/2017 3:52 PM, Jari Fredriksson wrote:
>
> I have had a VM in Google Cloud with 32 cores as a backup that my 
> raspberry automatically rises up and calls, if it can not reach my 
> Core i7 PC. And always on saturdays for net-jarif.
>
> Now Google announced that they have 96 cores available!
>
> I will now run 96 cores on weekdays (backup) and 192 threads for 
> net-jarif. This is amazing.
>
> And all orchestrated by a Raspberry Pi 2.
>
That... Is.... Awesome...

I just had a client that had a spin up systems on AWS with Skeddly to 
shutdown.  They missed the little checkbox on Skeddly that "this rule is 
not a test".

They hit a million boxes before AWS called them up and asked them if 
they were really trying to run a million boxes.  Don't be them :-)

Regards,
KAM