Attending: Ethan, Rob, JoshP, JoshC, Felix, RobBirnie
Short triage this week while working on 3.7 release.
New:
Facter
771: Fact-685 processor speed fact on OpenBSD
- Ethan to merge
770: PUP-3128 Added support for Manjaro Linux to Facter
- JoshP to notify contributor about contributor guidelines prior to merge
769: FACT-680
- Ethan to notify contributor to add test, update commit msg
Puppet:
2997: (PUP-3094) - Adding separate binaries for SELinux
- JoshP commenting that this seems like more of a packaging issue;
pinged Haus.
- JoshP to close ticket, start a thread on #puppet-dev google group
to discuss SELinux hindrances / implementation approaches
2986: Added yaml output to puppet resource #2986
- Pinged Henrik and Eric to get their opinions; Eric is discussing
it with contributor in the PR.
- If we move forward, would need a ticket and specs.
- Ticket has been created, some discussion has moved forward, it
seems like this is a useful operation. Still no tests, code requires
some additional cleanup.
Held Over
Facter:
748: FACT-639 Allow any kind of hypervisor for ec2 facts
- Kylo and Adrien are talking with the contributor
- Some concerns that the specs are incomplete (not checking enough
cases) Felix commenting
- Still under discussion as of 8/27
Puppet:
2866 PUP-1244 puppet4
- Switches the yum provider to use RPM version comparison instead of
semantic version comparison
- This might be breaking but is more correct and has been requested
since Puppet 0.24, document as a breaking change in Puppet 4
- Commit messages need to be fixed up, Andy to handle
- Has an error; @jantman will be looking into it this week; check
back next week.
- Still waiting on contributor for changes; Joshua to ping
- Jason has updated the failing case JoshP saw; Joshua to test again
- Tests out fine, and has been squashed, but seeing a failure on
1.8.7; JoshP will look at it
- As of 8/27, JoshP still investigating failure on 1.8.7
2952: PUP-3038 Cache search_directories
- Ongoing discussion about unexpected impact from this change
- Breaking specs
- Commit message neds updating to provide ticket # and information
about the change
- Felix also wondered how this related to PUP-1592 and has updated
PUP-3038 to link that
- Ethan added note that PUP-3032 might address this already, and
asked contributor to measure impact on his environment. It's possible
this ticket might be closed if the perf gains with the work on
PUP-3032 are sufficient.
2951: PUP-3038 Cache typeloader misses
- Same PUP-3038 ticket as 2952, similar concerns
- This is something that seems like a reasonable idea, but could
have some difficult to debug failure scenarios (when changing
environments for instance), that we're uncertain of at this point.
- Suggest we use a sentinel value for a TypeNotFound situation,
rather than relying on nil (even if it adds more code to the
PR, this will be more clear)
- We'd like to evaluate the concept for Puppet 4, but making this
part of 3.7 is not going to happen right now.
- JoshP to discuss Henrik comments about viability of this approach.
2919: Proposal for processing an external trusted OID yaml file
- JIRA issue has been filed for this pull request
- Needs spec tests, needs refining of the implementation. We'll
continue work on this with the contributor.
- Check in with Adrien tomorrow 8/14 about picking this up now that
he's rotated off of community role
- JoshP to check in on the settings behavior, given we don't want to
use a class variable to track init state (and have several different
hook options available for determining when settings can be changed)
2915: Set accept-encoding to identity when http_compression is false
- Looks fine; waiting for one spec name fix from Contributor, and
pinged Josh Cooper for any comments, as he was last to mention this
code change in the ticket.
- Josh commented on this indicating that the puppet module tool will
need a similar fix; we don't have enough information on this to
proceed, waiting for feedback from Andy.
- Need to look into how to handle this for puppet module tool as well
- It was discovered that zlib could be set to false in
https://github.com/puppetlabs/puppet/commit/1c69af2abea55579373e6111cf840197c7cb3494
based on ticket http://projects.puppetlabs.com/issues/2855
- Ethan to research if the Puppet[:zlib] setting could be deprecated
in 3.7 to be removed in Puppet 4 - Rubys http.rb already tries to load
zlib and fails when it's not present, so it might be good to remove
this setting from Puppet (might be able to piggyback on b64_zlib_yaml
ticket that exists)
- Ethan to pull this in, with additional change to PMT
lib/puppet/forge/repository and merge to master, once 3.7.0 has
shipped
2649: Add slot support for portage package provider.
- Duplicates GH-2779, decision is to ping both contributors to see
if they can coalesce their individual PRs into a single branch. Finch
commenting this week.
- GH-2779 hasn't seen updates, closing that one in favor of this
one. There's some concern about possibly using `:` as a special value
in package resource titles but we don't want to go down that route.
- Pinged Adrien for a final call on this now that the specs are updated
- Ethan to look at next week at some point
2683: Add more config manipulation options
- Needs a JIRA issue and tests are failing
- Would be a good fix to have, if contributor doesn't get back to us
might be worthwhile to fixup ourselves.
- Adrien asking for a Ticket so we can pull it onto community board
and tackle it.
- No response/no ticket, Joshua to create ticket
- Discussion that we should preserve the current behavior of `puppet
config set prop ''` and you should use `puppet config del` to remove
the setting and restore the default. JoshP to continue work.
- Comments from JoshP from 19 days ago have not been addressed as of 8/27
2770: Allow version and held states of a package to specified as
separate properties
- This is a very big and sticky issue, and might be worth promoting
into the platform team because of the wide ranging ramifications
- Plan is to discuss how we promote issues like this, and respond to
the contributor with our next steps forward
- Pending contributor response, willing to wait on this since we're
asking a lot of the contributor
- Will need to talk with Adrien tomorrow and figure out how to hand this off.
- Didn't have time to look at on 8/27 -- major PR that has breaking
changes and could require work. Would need to be targeted at Puppet 4
at the very least, but this one is tricky. Will try to investigate at
next PR meeting.
2870 (PUP-19708) Make the puppet agent retry failed requests.
- Issue in commit message references projects.puppetlabs.com ticket,
actual issue is https://tickets.puppetlabs.com/browse/PUP-2526
- There are a lot of cases where HTTP retries aren't meaningful
(catalog retrieval, report submission)
- We haven't been able to come to an agreement with the contributor
about what requests should be retried and how, and we might not have
the resources to make the changes ourselves.
- Felix commented in the ticket, some discussion is happening there
- Waiting to see if we can come to some agreement in the ticket as
to the direction of the change.
- As of 8/27, still stewing on this one
2869 Allow Upstart jobs on Amazon Linux
- Ticket filed, issues pending on the pull request
- Waiting on contributor feedback
- Needs a spec and an updated commit message
- Ethan pinged contributor on 8/27
- Ethan to verify again this is ready, and will merge to master once
3.7 has shipped
2820: (PUP-2867) Support ~ in file paths, allows to deploy to ~<user>
directories
- Already in discussion on puppet-dev
- Tests are failing, commits need squashing; waiting for contributor response
- What happens when the path starts with `~/` - do we expand to the
current user?
- Felix clarified the above issue; waiting on contributor
- As of 8/27, debate ongoing - will revisit in next PR triage.
2614: (PUP-2452) Refresh support for Package type
- Looks reasonable.
- There is a question from the contributor about if there needs to
be something changed to use this.
- The answer to the question is, "yes, having a use of this is ideal".
- Adrien noted that using the :methods system for the feature is
hard to control. Better is to require that the provider just
explicitly declare the feature, which it already does.
- Another concern came up about getting refresh events that will
cause unintended reinstalls. Such a case is when a class contains a
package and is the target of a notify.
- This is a larger change to functionality that we need to bring up
on puppet-dev to figure out how to deal with the unintended
consequences of adding this functionality.
- Adrien to file a ticket on finer grained refresh handling and
start a thread on puppet-dev to come to a decision about how to better
control refresh events.
- Some discussion taking place in the PR, waiting on the contributor
for an updated opt-in approach
- As of 8/27, contributor still working on PR - will re-eval next week
--
--
Ethan Brown
et...@puppetlabs.com
Senior Software Engineer
Join us at PuppetConf 2014, September 20-24 in San Francisco
Register by September 8th to take advantage of the Final Countdown —save $149!
--
You received this message because you are subscribed to the Google Groups
"Puppet Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to puppet-dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/puppet-dev/CALCHMcTijZPMM8JQ_y30TDHGfKbvSrobUYLbEjbhg1cbLfY6Vg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
