Re: [Vote] Retire 32 bit images
HVM being the preferred virt type in EC2, +1 for abandoning 32-bit arch. However, I'm biased in favor of Amazon's infrastructure. Cheers, milan 2015-06-17 11:28 GMT+02:00 Daniel P. Berrange berra...@redhat.com: On Wed, Jun 17, 2015 at 12:02:15AM -0400, David Gay wrote: For some time now, there's been discussion of ending production of 32 bit cloud images. Dusty's been kind enough to create a ticket regarding this matter[1]. Thoughts on this? +1/+0/-1? There could be a reason that we shouldn't stop producing 32 bit images that someone hasn't voiced yet. We can discuss during the meeting this week, as well. What usage statistics are available to backup an assertion that we can abandon creation of 32-bit images ? Do we have download stats for the different architecture images that show people aren't using 32-bit ? Regards, Daniel -- |: http://berrange.com -o-http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :| ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Fedora EC2 Amis: SriovNetSupport
That would be one way. However, since it actually is a majority of the instance types[1] that support this feature, I'd vote to make it the default. That's unless it prevents using/booting the small instances of course, what AFAIK isn't the case --- I was able to create a snapshot ami from an F21 instance, register it with the sriov flag and boot both a t2 and an r3 instances without any issue and confirmed the sriov flag was in effect on the r3 instance. The only limit I know of is the ami has to be HVM[2]. Moreover, the feature is for free, no extra charges apply so why not to take advantage of lower latencies on the instances. Cheers, milan [1] Enhanced Networking Support, The Instance Types Matrix, http://aws.amazon.com/ec2/instance-types/ [2] Enabling Enhanced Networking on Other Linux Distributions, http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking.html 2015-03-19 18:39 GMT+01:00 Garrett Holmstrom gho...@fedoraproject.org: On 2015-03-19 9:11, milanisko k wrote: I'd like to ask about the status/plan of SriovNet support[1] for Fedora Amazon images. I was able to set this up manually for recent F21 ami ami-5cd9ea41, but it is quite an inconvenient experience --- one has to instantiate, stop, set attribute value through the CLI tool and start again to enable the feature. Would it be possible to register future Fedora amis with the enhanced networking flag enabled[1]? As far as motivation is concerned, please check the blog post[2] for some performance evaluation. Since that only works for some types of instances it wouldn't make sense to do that for all images, but registering a separate image (from the same bundle/snapshot) with that enabled would be reasonable. -- Garrett Holmstrom ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Cloud image lifetimes
I vote for different life-time for different purpose, precisely as David suggests: scratch, test, RC, Final (Beta). But let's also don't forget about policy for testing the amis: how long to keep amis that failed testing in particular release stage. Ideally, we should also keep track of released/available amis so people are able to easily find particular version and flavor of Fedora amis. My usual use case is: reproduce a bug on e.g. F20 + some software stack. I find F20 ami ideal for this but the issue is to locate it[1]. Cheers, milan [1] http://thecloudmarket.com/ 2015-03-19 5:11 GMT+01:00 M. Edward (Ed) Borasky zn...@znmeb.net: On Wed, Mar 18, 2015 at 3:38 PM, David Gay d...@redhat.com wrote: Greetings! We sort of ran out of time in today's Cloud WG meeting, but I did want to ask: What are your thoughts on AMI lifetimes? That is to say, how long should EC2 AMIs exist before they're deleted? A few points to consider: - AMIs only cost us for storage, so it's not a *huge* cost to maintain a public AMI - At the same time, there are a lot of AMIs, since we build 2-4 per AWS region per build, and that number is growing - There are 9 regions now, and we have 2 virtualization types, and 2 volume types, as well (9 regions * 2 * 2 = 36 AMIs per Base image build, 18 for Atomic builds (since they are only available in HVM format)) - This total number will only grow larger as we add instance-store AMIs, and so on - This isn't even taking into account any costs we'll have once we secure a deal with other providers like HP, Rackspace, and GCE, to maintain public images on their services I propose we have some sort of discussion regarding how long cloud image builds should be available on services like AWS. I suspect this will resolve to having different lifetimes for scratch, test, RC, final, and maybe other build types. Any input is appreciated. We can certainly talk about this at next week's meeting, as well. -- David ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct In God We Trust - All Others Bring Data ;-) Surely you or someone on the team must have some raw data on usage for the existing AMIs, including comparisons for how much usage Fedora AMIs get vs. CentOS AMIs and RHEL AMIs. Don't build / maintain what people aren't using! If you have raw data, I'd be happy to explore it for you - email me off-list if you need an NDA or something like that. -- OSJourno: Robust Power Tools for Digital Journalists http://www.znmeb.mobi/stories/osjourno-robust-power-tools-for-digital-journalists Remember, if you're traveling to Bactria, Hump Day is Tuesday and Thursday. ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Fedora EC2 Amis: SriovNetSupport
Hello list, I'd like to ask about the status/plan of SriovNet support[1] for Fedora Amazon images. I was able to set this up manually for recent F21 ami ami-5cd9ea41, but it is quite an inconvenient experience --- one has to instantiate, stop, set attribute value through the CLI tool and start again to enable the feature. Would it be possible to register future Fedora amis with the enhanced networking flag enabled[1]? As far as motivation is concerned, please check the blog post[2] for some performance evaluation. Thanks, milan [1] http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking.html [2] http://blogs.scalablelogic.com/2014/01/enhanced-networking-in-aws-cloud-part-2.html ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: 32bit AMI issues (unofficial AMI)
looking at [2], lines 605--612, there's just the loop-back device therefore cloud-init fails to fetch instance metadata. Not sure why there's no other NIC device, though... -- milan 2014-10-30 16:55 GMT+01:00 Dusty Mabe du...@dustymabe.com: Last night I created some AMIs for roshi to test using the procedure at [1]. I then booted the x86_64 AMI (ami-145fdd7c) and it seemed fine. The i386 AMI (ami-de6be9b6) had issues booting (see [2] for full log). Has anyone else seen this before? [[32m OK [0m] Started System Logging Service. [[32m OK [0m] Started OpenSSH Server Key Generation. [ 17.320770] cloud-init[332]: Cloud-init v. 0.7.5 running 'init-local' at Thu, 30 Oct 2014 04:30:06 +. Up 17.22 seconds. [[32m OK [0m] Started Initial cloud-init job (pre-networking). [[1;31mFAILED[0m] Failed to start LSB: Bring up/down networking. See systemctl status network.service for details. Thanks, Dusty [1] - https://gist.github.com/dustymabe/3d53cfe6b9ae32c5e2fb [2] - http://ur1.ca/im3op ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Cloud Test Overview
Fair enough. Not sure about the Taskotron features (plan) but our harness already supports couple of things which are a must for the load we've been facing wr. images validation: - parallel instances handling - parallel test cases execution - test case dependencies - remote (bash) code execution (rpyc) - expect-like test support (paramiko + bunch of handy methods) - dry-run mode - event-driven (gevent) - capable of handling ~200 simultaneous connections from a single process - custom testing modules support - test case tags, stages, filters and applicability handling - instance reboot handling - plain YAML result files, easily to translate to e.g. xunit - GPLd, GitHub-hosted, open-source These give us a workhorse for which we're missing a fedimg listener atm. Actually, we have to implement one anyway because of our release engineering adopting a process similar to fedmsg. Sorry for the advertisement ;) but we're missing a community and should the community miss a harness, there's one worth of considering I think... Cheers, milan 2014-10-08 1:21 GMT+02:00 Mike Ruckman ro...@fedoraproject.org: On Fri, Oct 03, 2014 at 01:57:43PM +0200, milanisko k wrote: Guys, are the Cloud Image news being published on the Fedmsg channels? I'd like to investigate integration of RedHatQE's test runner and suite https://github.com/RedHatQE/dva. Thanks, milan Milan, Currently fedimg has some smoketests in it, but the plan for the future is to decouple the tests from fedimg and move the testing to Taskotron. We're still a ways out before that can happen, but it's the end goal we're looking to accomplish. -- // Mike -- Fedora QA freenode: roshi http://roshi.fedorapeople.org ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Cloud Test Overview
Thanks, seems fedimg is going to be both an image upload and test service, right? Cheers, milan 2014-10-03 14:22 GMT+02:00 Matthew Miller mat...@fedoraproject.org: On Fri, Oct 03, 2014 at 01:57:43PM +0200, milanisko k wrote: are the Cloud Image news being published on the Fedmsg channels? I'd like to investigate integration of RedHatQE's test runner and suite https://github.com/RedHatQE/dva. Amazingly, yes -- that plan is slowly, slowly coming to fruition. Take a look at http://fedimg.readthedocs.org/en/latest/ for the Fedimg service, and particularly http://fedimg.readthedocs.org/en/latest/consumer/ http://fedimg.readthedocs.org/en/latest/messaging/ -- Matthew Miller mat...@fedoraproject.org Fedora Project Leader ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Cloud Test Overview
Cool, thanks. Smoke-testing at this step is a reasonable thing to do. I'm asking because our RCM is currently in the process of adopting a strategy similar to fedimg so I thought we could donate couple of extra cycles should the community benefit from our test harness. Anyways, we'd be delighted to share. Cheers, milan 2014-10-07 17:47 GMT+02:00 Matthew Miller mat...@fedoraproject.org: On Tue, Oct 07, 2014 at 05:28:27PM +0200, milanisko k wrote: seems fedimg is going to be both an image upload and test service, right? The idea is for it to do some really simple initial smoketests, to save further testing systems from wasting time. I don't think we want it to grow to become yet another test harness. -- Matthew Miller mat...@fedoraproject.org Fedora Project Leader ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Cloud Test Overview
Guys, are the Cloud Image news being published on the Fedmsg channels? I'd like to investigate integration of RedHatQE's test runner and suite https://github.com/RedHatQE/dva. Thanks, milan 2014-10-03 3:01 GMT+02:00 Mike Ruckman ro...@fedoraproject.org: Hey fellow fedorans! I've made some slight edits to the existing Test Overview [0] on the wiki and would like some feedback. It's meant as a high-level overview of all the things we think we need to test for the Fedora 21 cloud product. It's not intended to be an exhaustive list, just a basic statement of intent. So if you have a couple minutes to look it over (it's short) and look for anything I might have missed, it would be greatly appreciated. Thanks! [0] https://fedoraproject.org/wiki/User:Roshi/QA/Cloud_Docs/Test_Overview -- // Mike -- Fedora QA freenode: roshi http://roshi.fedorapeople.org ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: help needed (soooo much help needed)
@gholms, hi, what are the plans then? ;) Thanks! milan ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: help needed (soooo much help needed)
Hi list, Matt, I'm looking at https://fedorahosted.org/cloud/ticket/23; What is the desired status of the refactoring? Have some changes already been agreed upon? Would it be possible to replace cloud-init completely? (For instance, https://github.com/cernvm/amiconfig looks quite minimalistic...) Cheers, milan 2014-04-03 20:37 GMT+02:00 Matthew Miller mat...@fedoraproject.org: https://fedorahosted.org/cloud/query?status=acceptedstatus=assignedstatus=newstatus=reopenedsummary= ^Filemilestone=Fedora+21+%28Feature+Deadline%29order=prioritycol=idcol=summarycol=milestonecol=statuscol=ownercol=component ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: unavailable F20 amis
Hi, it seems OK save for Sao Paulo 32 bit: ami-6f6ecf72 Thanks, milan 2013/12/18 Matthew Miller mat...@fedoraproject.org On Wed, Dec 18, 2013 at 03:12:02PM +0100, milanisko k wrote: following AMIs can't be instantiated --- AWS error message says that these either have been deleted or I've got no access (but US East 1 amis works for me): This should be fixed now -- very sorry about that! -- Matthew Miller -- Fedora Project Architect -- mat...@fedoraproject.org ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
why (not) update amis
Yesterday I debugged an installation of a pulp-server F19 node in EC2. Of course, I hit a Selinux issue[1] which could have been solved by a simple 'yum update -y' of the affected package. The thing is everyone would have to do that. Given the fact ami instances are ephemeral/disposable single-shot entities appearing and disappearing in huge numbers with significant frequency world-wide, such a dumb thing like updating the system right after boot has negative impacts on many things starting from ones EC2 budget, continuing through wasting fedora content bandwidth and ending with Carbon Dioxide pollution stemming in the CPU cycles wasted and all of that just because the amis aren't up-to-date... Let's preserve the planet for our children by more frequent ami respins! -- milan [1] https://bugzilla.redhat.com/show_bug.cgi?id=877831 ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
yum scriptlets fail in ec2 userdata script
Hey guys, I'm hitting this issue[1] all the time with a userdata script [2]. Based on the error code, sh complains about non-existing command. My suspicion is some filesystem service isn't running in the time userdata scripts are evaluated by systemd/cloud-init. I'm unable to locate the issue, though... Thanks for help, milan [1] http://paste.fedoraproject.org/41006/ [2] https://github.com/RedHatQE/pulp-automation/blob/master/ec2deploy/fedora_pulp.sh ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: yum scriptlets fail in ec2 userdata script
done: https://bugzilla.redhat.com/show_bug.cgi?id=1010335 -- milan 2013/9/20 Matthew Miller mat...@fedoraproject.org: On Fri, Sep 20, 2013 at 03:45:58PM +0200, milanisko k wrote: I'm hitting this issue[1] all the time with a userdata script [2]. Based on the error code, sh complains about non-existing command. My suspicion is some filesystem service isn't running in the time userdata scripts are evaluated by systemd/cloud-init. I'm unable to locate the issue, though... Can you file this as a bug against selinux-policy-targeted? That might not be right in the end but will get the issue tracked. -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct ___ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct