Can i keep foreman from including veth* interfaces? causes performance issues

Running puppet on my mesos slaves that have a ton of containers
coming/going and foreman is barfing on handling all of the veth* interfaces
that come and go. Is there any way to keep this from happening? for example
if you login to foreman dashboard and view interfaces on the slaves often
times chrome will crash and then i get alerts that all of the CPU's on the
foreman web node are pegged.

You can set ignore_puppet_facts_for_provisioning to true or wait for 1.10
to be released and use ignored_interface_identifiers.

··· On Tuesday, December 22, 2015 at 9:35:40 AM UTC-8, Byron Miller wrote: > > Running puppet on my mesos slaves that have a ton of containers > coming/going and foreman is barfing on handling all of the veth* interfaces > that come and go. Is there any way to keep this from happening? for example > if you login to foreman dashboard and view interfaces on the slaves often > times chrome will crash and then i get alerts that all of the CPU's on the > foreman web node are pegged. >

You can set ignore_puppet_facts_for_provisioning to true or wait for 1.10
to be released and use ignored_interface_identifiers.

··· On Tuesday, December 22, 2015 at 9:35:40 AM UTC-8, Byron Miller wrote: > > Running puppet on my mesos slaves that have a ton of containers > coming/going and foreman is barfing on handling all of the veth* interfaces > that come and go. Is there any way to keep this from happening? for example > if you login to foreman dashboard and view interfaces on the slaves often > times chrome will crash and then i get alerts that all of the CPU's on the > foreman web node are pegged. >

Yep, but only in 1.10 (see [1] for the code) - there's a
Setting[:ignored_interface_identifiers] which is a regexp of devices to
ignore. Should do the job.

Until that comes out, you can hack the equivalent method in 1.9 [2] to add
you devices to the hardcoded regexp

Greg

[1]


[2]

··· On 22 December 2015 at 17:35, Byron Miller wrote:

Running puppet on my mesos slaves that have a ton of containers
coming/going and foreman is barfing on handling all of the veth* interfaces
that come and go. Is there any way to keep this from happening? for example
if you login to foreman dashboard and view interfaces on the slaves often
times chrome will crash and then i get alerts that all of the CPU’s on the
foreman web node are pegged.

Will this second method for 1.9 work if discovery plugin is enabled?

··· On Tuesday, December 22, 2015 at 3:09:33 PM UTC-8, Greg Sutcliffe wrote: > > On 22 December 2015 at 17:35, Byron Miller <byr...@gmail.com > > wrote: > >> Running puppet on my mesos slaves that have a ton of containers >> coming/going and foreman is barfing on handling all of the veth* interfaces >> that come and go. Is there any way to keep this from happening? for example >> if you login to foreman dashboard and view interfaces on the slaves often >> times chrome will crash and then i get alerts that all of the CPU's on the >> foreman web node are pegged. >> >> > Yep, but only in 1.10 (see [1] for the code) - there's a > Setting[:ignored_interface_identifiers] which is a regexp of devices to > ignore. Should do the job. > > Until that comes out, you can hack the equivalent method in 1.9 [2] to add > you devices to the hardcoded regexp > > Greg > > [1] > https://github.com/theforeman/foreman/blob/develop/app/services/fact_parser.rb#L160-L162 > [2] > https://github.com/theforeman/foreman/blob/1.9-stable/app/services/fact_parser.rb#L165-L167 >