Re: [Spacewalk-devel] On quality of patches
On 3/4/14, 12:31 AM, Silvio Moioli smoi...@suse.de wrote: From: Paul Robert Marino prmari...@gmail.com mailto:prmari...@gmail.com I think we need a map of what these shared file and functions effect so we can do more formal QA testing in the future On 03/03/2014 08:09 PM, Lamont Peterson wrote: Agreed. Unfortunately I think such a map, if ever existed, would be far too complex to be of any practical value. Consider this image, which maps only database tables, a relatively small fraction of the total software complexity: http://turing.suse.de/~smoioli/relationships.real.compact.png Developers already have tools in modern IDEs such as Eclipse to check for (partial) call trees[1], and are expected to use them. I do not think we can do much more than that in this regard, of course new opinions are welcome :-) Cheers, [1] http://eclipse-tools.sourceforge.net/call-hierarchy/usage.html -- Silvio Moioli SUSE LINUX Products GmbH Maxfeldstraße 5, 90409 Nürnberg Germany ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel You’re right (and thanks for the vivid DB relationships diagram). In my mind, the auto-generated map should simply be a calling-me list: Function A is called by: file.java : 45 file2.java : 134 That’s for developers. The map for QA could be simpler and generated from merely stuff one can hit “submit” on? -- Lamont Peterson Sr. Systems Administrator | Unix Systems Operations Intermountain Healthcare Office: 801.442.6497 | lamont.peter...@imail.org ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel
Re: [Spacewalk-devel] Ubuntu Sync Script
Hi, Martin. I would suggest splitting out such functionality. The spacewalk-repo-sync functionality would be exactly as it is today, simply add support to use it for Channels with .deb Repositories. The other functionality that you mention might be a new script or some parts of it might make more sense to be covered in the pre-defined channel sets. -- Lamont Peterson Sr. Systems Administrator | Unix Systems Operations Office: 801.442.6497 [cid:image002.jpg@01CEC98D.A33033E0]http://intermountainhealthcare.org/ From: Martin Juhl m...@casalogic.dkmailto:m...@casalogic.dk Reply-To: spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com Date: Monday, January 20, 2014 at 1:48 AM To: spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com Subject: Re: [Spacewalk-devel] Ubuntu Sync Script Hi Michael (and all others) I have close to no experience writing python code.. ( I might learn in the future), so I hope that someone else will take the job of rewriting this script... Besides of that, I'm not sure that all of the functionality of the script can be merged into spacewalk-repo-sync.. my script also contains functionality to check which distributions and sub-channels are available from Ubuntu.. Is there anything else I can do to help people rewrite/use this script for now??? Regards Martin Fra: Michael Mraka michael.mr...@redhat.commailto:michael.mr...@redhat.com Til: spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com Sendt: torsdag, 9. januar 2014 16:00:31 Emne: Re: [Spacewalk-devel] Ubuntu Sync Script Hello Martin, % I have created a bash script for syncing Ubuntu Channels to SpaceWalk.. % % The script has been written to resemble the satellite-sync and mgr-ncc-sync on Red Hat Satellite and SuSE Manager.. % % The script is definitely NOT perfect, and could do with a rewrite in python/perl.. But it works, and thats the main reason I'm releasing it... Your sync script looks good. If you really think about rewriting it into python I'd point you to spacewalk-repo-sync command. It is currently able to download packages only from yum repos but you can extend it with new repo plugin (see ./backend/satellite_tools/repo_plugins in spacewalk.git). So you have to write code only for fetching package list and packages itself and rest of the infrastructure is already there. Regards, -- Michael Mráka Satellite Engineering, Red Hat ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.commailto:Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel inline: 43D9F316-7150-4FD9-97C0-94E6B1732B1F[94].png___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel
Re: [Spacewalk-devel] On quality of patches
On 1/30/14, 3:57 AM, Matej Kollar mkol...@redhat.com wrote: Hi all. We welcome community contributions and we want to maintain some level of quality. Natural expectation is that every proposed patch is tested for the functionality. However, recently I came across quite nasty little thing in Spacewalk... Using SSM to schedule reboot one might end up very unpleasantly surprised, expecting it not to occur for specified period of time, when in fact it is being scheduled as soon as possible, no matter what you try. (There was no way it ever worked - parameters for forms were not passed to script in request (where DatePicker looks for them) and new Date object was returned. Well, someone scamped his work, so I had to do it for him last night. And I consider myself lucky that I found out... As en exercise in applied imagination, try to imagine OSAD running on such machine. To stretch your imagination even further, imagine the machine acting as a life supporting computer during child surgery...) Nice example :) . FYI, around here (Intermountain Healthcare), any activity requiring a planned reboot of a server would also require an SA to be on “live” and monitoring the system and would likely involve a DBA or App Admin to be online to confirm all services are properly started/running. Therefore, this example, while great, wouldn’t happen in our industry because we would never use such functionality (yes, mistakes can happen). BTW, there are systems that can directly impact patient care in general bed space, ICU, NICU, ED, OR and other facilities. So this example is an excellent idea in general to keep in mind. Yes, most systems in the world will not cause harm to life, limb or property should something go wrong, but there are a lot of places where it could (not just health care, but think about Air Traffic Control, City-wide Traffic Management, National Weather Service, Automated Industrial/Manufacturing and the list could go on and on). In the other end of the scale, what if something went sideways in a small company? People with 10 servers to manage or 100 servers and 5 employees can go out of business if their stuff is down/broken (by systems management software, like Spacewalk?). That’s a huge impact to their worlds. Btw: next time have a peek into struts-config.xml and look for form-bean and form-property tags... say like in eea3c6320bfc3dae8532844a7c0e71dcc5712c39. To continue with our story: Purpose of that form is to schedule action to occur not sooner than. It fails on all but one particular case. It simply cold not have been tested. Reboot is considered destructive operation. Therefore we insisted on two-step scheduling for this feature. I fixed this one but wonder how many other similar features there were recently introduced. And the final rhetorical question: Have the author even tried it before submitting the patch? I tend to doubt it. Matej ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel Thanks! Keep up the great work. -- Lamont Peterson Sr. Systems Administrator | Unix Systems Operations Intermountain Healthcare Office: 801.442.6497 ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel
Re: [Spacewalk-devel] On quality of patches
Agreed. Such a map needs to be generated from source, please? -- Lamont Peterson Sr. Systems Administrator | Unix Systems Operations [cid:image002.jpg@01CEC98D.A33033E0]http://intermountainhealthcare.org/ From: Paul Robert Marino prmari...@gmail.commailto:prmari...@gmail.com Reply-To: spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com Date: Thursday, January 30, 2014 at 9:50 AM To: spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com spacewalk-devel@redhat.commailto:spacewalk-devel@redhat.com Subject: Re: [Spacewalk-devel] On quality of patches There is an other factor here that no one has mentioned. Recently I've found and reported a lot of bugs in the nightly repo. Where they came from is usually some one updates a portion of the code to fix a specific issue but since a portion of that code is called by multiple parts of the web interface it may break other pages than the one the developer is focused on. I think we need a map of what these shared file and functions effect so we can do more formal QA testing in the future -- Sent from my HP Pre3 On Jan 30, 2014 7:33, Duncan Mac-Vicar P. dmacvi...@suse.demailto:dmacvi...@suse.de wrote: On 30/01/14 11:57, Matej Kollar wrote: Hi all. We welcome community contributions and we want to maintain some level of quality. Natural expectation is that every proposed patch is tested for the functionality. I fully agree with you that this is not acceptable. Now, I think your diagnostic is IMHO neither fair or correct. This does not have to do with the community contributions but with the setup of the project itself. We sometimes have sent patches that are very strictly reviewed, sometimes needing change multiple times to get a reviewer happy. Then next day our internal testsuite fails only to realize that someone with direct commit access did a big refactoring and committed very broken code in a big patch that was not reviewed by anyone and which quality was also obviously not the best. We asked ourselves, what is the point of reviewing some of them?. Not only code but also design decisions and way of approaching solutions. Sometimes this happens with our own patches, depending on the reviewer, they may get committed faster. The setup of the review process is broken. - All code should be committed in similar units (features/branches) - All code should be able to be reviewed and vetoed by everyone OpenStack has this model working quite successfully. Every patch is reviewed with +1, and they need a certain amounts of ACKS to get committed. Everyone can review and people learn in the process, and it is a great source of inspiration for other projects. We care about quality and you can see that we not only contributed our own testsuite in early 2011 with the first release of our own product, but then focused lot of contributions in having the testsuites enabled and working again. But all this is pointless if the process has holes. Right now it depends on a lot of luck to be useful. We suggested not long time ago in the mailing list to move to a github approach where code could be submitted only with pull-requests that need to be reviewed. Policies could be created that at least 2-3 acks is needed to merge a feature. People can study those reviews and learn in the process. Continuous integration could be setup on top of this process, to have up to date results. But what you are seeing, the process is kind of designed to produce those results, which is a process where code has different ways to arrive to master, which different quality outcomes. -- Duncan Mac-Vicar P. - http://www.suse.com/ SUSE LINUX Products GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg) Maxfeldstraße 5, 90409 Nürnberg, Germany ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.commailto:Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel inline: 43D9F316-7150-4FD9-97C0-94E6B1732B1F[96].png___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel
Re: [Spacewalk-devel] upstreaming some of our .spec file changes
[snip] Well, discussions about renaming all tools and packages to spacewalk- something pop up from time to time for many years ;). Althought we mostly agree with renaming comandline tools (assuming that old names will remain at least as symlinks for compatibility reasons) we've decided not to rename rpm packages because it will make upgrades much complicated. How so? You simply Obsoletes: the old package name with the new one. You can even drop the Obsoletes: from the SPEC file after a time. [snip] -- Lamont Peterson lamont.peter...@imail.org Senior Systems Administrator Intermountain Healthcare ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel
[Spacewalk-devel] Spacewalk web interface SSL certificate instructions?
All, It seems that the only good(?) instructions for how to update or replace the SSL certificate used for the Spacewalk web interface were at a site whose URL I'd prefer to not fully include or quote, but, here goes: [ http://unf***ablelinux.com/2008/07/02/spacewalk-and-avoiding-self-signed-certificates/ ] (which presented problems even trying to visit though our company proxies). However, it now seems that this site is just gone (I tried it at home). Does anyone have a copy of those instructions or something better or some other site? Once I have this task done, I'd be happy to write it up for the Spacewalk Documentation site. I'm sure there are plenty of other people who want/need to accomplish this task besides me. -- Lamont Peterson lamont.peter...@imail.org Intermountain Healthcare, Information Systems ___ Spacewalk-devel mailing list Spacewalk-devel@redhat.com https://www.redhat.com/mailman/listinfo/spacewalk-devel