Re: [scl.org] Idea: Software Collections Daemons Made System-wide
Thanks for your input, guys, really appreciated. Honza On 03/22/2017 03:51 PM, Davis, Daniel (NIH/NLM) [C] wrote: Amen to this. I shipped an appliance install base on Pungi/Anaconda, but in my current role, I do not have root. I found SCL and got what we needed without us having to build it ourself. -Original Message- From: sclorg-boun...@redhat.com [mailto:sclorg-boun...@redhat.com] On Behalf Of Griffin, Wesley (Fed) Sent: Wednesday, March 22, 2017 10:49 AM To: Honza Horak; sclorg@redhat.com Subject: Re: [scl.org] Idea: Software Collections Daemons Made System-wide My use case is for newer compilers and interpreters on development workstations. I do not, however, manage my systems - our system administrator does that, so I work with him to get the SCL packages installed where necessary. I am pretty confident that if I wanted to install an SCL package and it wrapped the system binary in some way he would refuse. He supports many more machines than just my development group and tries to maintain the same set of packages across all machines. The beauty of SCL is that he can install it everywhere, but not affect anybody that doesn't opt-in. We have a similar setup for Intel compilers, Matlab, and other scientific software used in our organization. So in that sense, adapting to a new approach would be troublesome, but only if it was the only approach. As long as the system binary wrappers are in subpackages which are optional, then I see no issues for us to continue using future SCL packages. Thanks, Wes From: Honza Horak Sent: Wednesday, March 22, 2017 3:41:54 AM To: Griffin, Wesley (Fed); sclorg@redhat.com Subject: Re: [scl.org] Idea: Software Collections Daemons Made System-wide I don't expect we'll do this for existing packages we provide, it's more a vision for new collections (e.g. mariadb 10.2 at some point). Anyway, do you also see the same point for those upcoming collections (where we're not tight by keeping backward compatibility), e.g. because you're fine with all the SCL specifics and adapting to the new way would be troublesome? Honza On 03/22/2017 04:30 AM, Griffin, Wesley (Fed) wrote: I will chime in as the "old curmudgeon" - as long as you don't break the existing use case, then I'm fine with any changes. The blog post says subpackages will be used to ensure no breakage - I just want to re-iterate that this is important and should not get lost if the proposal changes. Thanks, Wes From: sclorg-boun...@redhat.com on behalf of Honza Horak Sent: Tuesday, March 21, 2017 11:05:26 AM To: sclorg@redhat.com Subject: [scl.org] Idea: Software Collections Daemons Made System-wide This is basically a kick-off for getting more feedback for an idea shared at http://www.themindiseverything.eu/2017/03/software-collections-daemons-made.html. Shortly, SCL has worked nicely for several years and people love them. But even the beloved ones have some issues. And what we hear from users, the issues with Software Collections concept currently are basically those: * we need to use scl enable, which changes $PATH and other environment variables, so the binaries placed in different location are visible by the system * scripts originally written for "normal" MySQL use full paths (like /usr/bin/mysql) which does not work when we only have the Software Collection installed * Data directory, config files and log files are on different location than it is common The blog post tries to summarize possible solution, which I'm looking for feedback now, ideally by replying to this mail.. TIA, Honza ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Idea: Software Collections Daemons Made System-wide
Amen to this. I shipped an appliance install base on Pungi/Anaconda, but in my current role, I do not have root. I found SCL and got what we needed without us having to build it ourself. -Original Message- From: sclorg-boun...@redhat.com [mailto:sclorg-boun...@redhat.com] On Behalf Of Griffin, Wesley (Fed) Sent: Wednesday, March 22, 2017 10:49 AM To: Honza Horak; sclorg@redhat.com Subject: Re: [scl.org] Idea: Software Collections Daemons Made System-wide My use case is for newer compilers and interpreters on development workstations. I do not, however, manage my systems - our system administrator does that, so I work with him to get the SCL packages installed where necessary. I am pretty confident that if I wanted to install an SCL package and it wrapped the system binary in some way he would refuse. He supports many more machines than just my development group and tries to maintain the same set of packages across all machines. The beauty of SCL is that he can install it everywhere, but not affect anybody that doesn't opt-in. We have a similar setup for Intel compilers, Matlab, and other scientific software used in our organization. So in that sense, adapting to a new approach would be troublesome, but only if it was the only approach. As long as the system binary wrappers are in subpackages which are optional, then I see no issues for us to continue using future SCL packages. Thanks, Wes From: Honza Horak Sent: Wednesday, March 22, 2017 3:41:54 AM To: Griffin, Wesley (Fed); sclorg@redhat.com Subject: Re: [scl.org] Idea: Software Collections Daemons Made System-wide I don't expect we'll do this for existing packages we provide, it's more a vision for new collections (e.g. mariadb 10.2 at some point). Anyway, do you also see the same point for those upcoming collections (where we're not tight by keeping backward compatibility), e.g. because you're fine with all the SCL specifics and adapting to the new way would be troublesome? Honza On 03/22/2017 04:30 AM, Griffin, Wesley (Fed) wrote: > I will chime in as the "old curmudgeon" - as long as you don't break the > existing use case, then I'm fine with any changes. The blog post says > subpackages will be used to ensure no breakage - I just want to re-iterate > that this is important and should not get lost if the proposal changes. > > Thanks, > Wes > > > From: sclorg-boun...@redhat.com on behalf > of Honza Horak > Sent: Tuesday, March 21, 2017 11:05:26 AM > To: sclorg@redhat.com > Subject: [scl.org] Idea: Software Collections Daemons Made System-wide > > This is basically a kick-off for getting more feedback for an idea > shared at > http://www.themindiseverything.eu/2017/03/software-collections-daemons-made.html. > > Shortly, SCL has worked nicely for several years and people love them. > But even the beloved ones have some issues. And what we hear from > users, the issues with Software Collections concept currently are basically > those: > > * we need to use scl enable, which changes $PATH and other environment > variables, so the binaries placed in different location are visible by > the system > * scripts originally written for "normal" MySQL use full paths (like > /usr/bin/mysql) which does not work when we only have the Software > Collection installed > * Data directory, config files and log files are on different location > than it is common > > The blog post tries to summarize possible solution, which I'm looking > for feedback now, ideally by replying to this mail.. > > TIA, > Honza > > ___ > SCLorg mailing list > SCLorg@redhat.com > https://www.redhat.com/mailman/listinfo/sclorg > > ___ > SCLorg mailing list > SCLorg@redhat.com > https://www.redhat.com/mailman/listinfo/sclorg > > ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Idea: Software Collections Daemons Made System-wide
My use case is for newer compilers and interpreters on development workstations. I do not, however, manage my systems - our system administrator does that, so I work with him to get the SCL packages installed where necessary. I am pretty confident that if I wanted to install an SCL package and it wrapped the system binary in some way he would refuse. He supports many more machines than just my development group and tries to maintain the same set of packages across all machines. The beauty of SCL is that he can install it everywhere, but not affect anybody that doesn't opt-in. We have a similar setup for Intel compilers, Matlab, and other scientific software used in our organization. So in that sense, adapting to a new approach would be troublesome, but only if it was the only approach. As long as the system binary wrappers are in subpackages which are optional, then I see no issues for us to continue using future SCL packages. Thanks, Wes From: Honza HorakSent: Wednesday, March 22, 2017 3:41:54 AM To: Griffin, Wesley (Fed); sclorg@redhat.com Subject: Re: [scl.org] Idea: Software Collections Daemons Made System-wide I don't expect we'll do this for existing packages we provide, it's more a vision for new collections (e.g. mariadb 10.2 at some point). Anyway, do you also see the same point for those upcoming collections (where we're not tight by keeping backward compatibility), e.g. because you're fine with all the SCL specifics and adapting to the new way would be troublesome? Honza On 03/22/2017 04:30 AM, Griffin, Wesley (Fed) wrote: > I will chime in as the "old curmudgeon" - as long as you don't break the > existing use case, then I'm fine with any changes. The blog post says > subpackages will be used to ensure no breakage - I just want to re-iterate > that this is important and should not get lost if the proposal changes. > > Thanks, > Wes > > > From: sclorg-boun...@redhat.com on behalf of > Honza Horak > Sent: Tuesday, March 21, 2017 11:05:26 AM > To: sclorg@redhat.com > Subject: [scl.org] Idea: Software Collections Daemons Made System-wide > > This is basically a kick-off for getting more feedback for an idea > shared at > http://www.themindiseverything.eu/2017/03/software-collections-daemons-made.html. > > Shortly, SCL has worked nicely for several years and people love them. > But even the beloved ones have some issues. And what we hear from users, > the issues with Software Collections concept currently are basically those: > > * we need to use scl enable, which changes $PATH and other environment > variables, so the binaries placed in different location are visible by > the system > * scripts originally written for "normal" MySQL use full paths (like > /usr/bin/mysql) which does not work when we only have the Software > Collection installed > * Data directory, config files and log files are on different location > than it is common > > The blog post tries to summarize possible solution, which I'm looking > for feedback now, ideally by replying to this mail.. > > TIA, > Honza > > ___ > SCLorg mailing list > SCLorg@redhat.com > https://www.redhat.com/mailman/listinfo/sclorg > > ___ > SCLorg mailing list > SCLorg@redhat.com > https://www.redhat.com/mailman/listinfo/sclorg > > ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Idea: Software Collections Daemons Made System-wide
Le 21/03/2017 à 16:05, Honza Horak a écrit : > This is basically a kick-off for getting more feedback for an idea > shared at > http://www.themindiseverything.eu/2017/03/software-collections-daemons-made.html. Having an -system-wide package, providing a set of symlinks looks like a nice idea This can become hard to create, when there is multiple packages providing commands / service: e.g php-cli => /usr/bin/php php-devel => /usr/bin/phpize (and some others) php-pear => /usr/bin/pear (and some others) php-fpm => /usr/bin/php-fpm and php-fpm.service php-pecl-xdebug => /usr/bin/debugclient Another possible tricky bit is the socket used For now php-fpm listen on a network socket (localhost:9000) in base system and SCL, so the SCL provided service can really simply replace the system default one (like for databases). But, if we switch to UDS in the future this will become a bit more tricky (each SCL listen on a different UDS), symlink "could" work (this need to be checked). BTW a bit more simple for PHP which doesn't need any wrapper ;) (a simple symlink to the SCL binary is enough) Just few other comments which may help (a bit out-of-topic with your proposal, but related to SCL usability) - have you tried the service alias ? (exists in both SysV and systemd) - moving man page to base system As it could be consider a bit strange to have to enable the collection to read the documentation explaining how to enable it... - providing the module configuration file Because "scl enable" is not the best user friendly command, and may raise PATH overflow when used a lot, "module load" seems better (at least to me) This is default with scl-utils v2, but works perfectly with scl-utils v1, just dropping the file in the right place (/usr/share/Modules/modulefiles/) Remi. ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg
Re: [scl.org] Idea: Software Collections Daemons Made System-wide
On Wed, Mar 22, 2017 at 1:05 AM, Honza Horakwrote: > This is basically a kick-off for getting more feedback for an idea shared at > http://www.themindiseverything.eu/2017/03/software-collections-daemons-made.html. > > Shortly, SCL has worked nicely for several years and people love them. But > even the beloved ones have some issues. And what we hear from users, the > issues with Software Collections concept currently are basically those: > > * we need to use scl enable, which changes $PATH and other environment > variables, so the binaries placed in different location are visible by the > system > * scripts originally written for "normal" MySQL use full paths (like > /usr/bin/mysql) which does not work when we only have the Software > Collection installed > * Data directory, config files and log files are on different location than > it is common > > The blog post tries to summarize possible solution, which I'm looking for > feedback now, ideally by replying to this mail.. If you change the exec line in the wrapper script to be: exec -a $0 `basename "$0"` "$@" I believe this may actually also deal with the Python sys.executable and hence setuptools and pipsi compatibility problems that I raised in https://bugzilla.redhat.com/show_bug.cgi?id=1385471 Otherwise sys.executable will still be set to the unwrapped binary, and any generated shebang lines would refer to that rather than the wrapper script. Cheers, Nick. -- Nick Coghlan Red Hat Platform Engineering, Brisbane ___ SCLorg mailing list SCLorg@redhat.com https://www.redhat.com/mailman/listinfo/sclorg