[SUSPECTED SPAM] Re: [gentoo-user] Blacklist one of the pool's rsync server?
On Wed, Feb 28, 2018 at 01:25:27AM +, Grant Edwards wrote > Is there any way for me to "blacklist" a pool rsync server so that > emerge --sync won't try to use it? How about an iptables rule to block the bad address? BTW... [d531][waltdnes][~] nslookup rsync.us.gentoo.org Server: 192.168.123.254 Address:192.168.123.254#53 Non-authoritative answer: Name: rsync.us.gentoo.org Address: 128.61.111.10 Name: rsync.us.gentoo.org Address: 128.61.111.8 Name: rsync.us.gentoo.org Address: 208.100.4.53 Name: rsync.us.gentoo.org Address: 156.56.247.193 Name: rsync.us.gentoo.org Address: 216.165.129.134 Name: rsync.us.gentoo.org Address: 141.219.155.230 Name: rsync.us.gentoo.org Address: 129.21.171.72 Name: rsync.us.gentoo.org Address: 134.161.116.17 Name: rsync.us.gentoo.org Address: 128.61.111.7 Name: rsync.us.gentoo.org Address: 128.61.111.9 Name: rsync.us.gentoo.org Address: 2607:f128:1:3::2 Name: rsync.us.gentoo.org Address: 2620:8d:8000:15:225:90ff:fefd:344c Name: rsync.us.gentoo.org Address: 2001:48a8:11:97d::84:84 An easy solution is to hard-code one working IP address in /etc/hosts as rsync.us.gentoo.org /etc/portage/repos.conf/gentoo.conf has the "sync-uri". Is it legal to specify something like... sync-uri = rsync://128.61.110.10/gentoo-portage -- Walter DnesI don't run "desktop environments"; I run useful applications
[gentoo-user] Re: Blacklist one of the pool's rsync server?
On 2018-02-28, Dalewrote: > Grant Edwards wrote: >> Is there any way for me to "blacklist" a pool rsync server so that >> emerge --sync won't try to use it? >> >> I'm using the sync-url rsync://rsync.us.gentoo.org/gentoo-portage, but >> one of the pool's servers is barely usable for me. I don't know if >> it's a server problem or if traffic between my and that server is >> routed through a Mars orbiter. When any other server is chosen, the >> file list scrolls by faster than you can read it. Via the Mars >> orbibiter the list ticks by about one file every 5-10 seconds. It >> will usually finish (eventually), but sometimes hangs and times out. >> It's been like for at least a year or two. >> >> [FWIW, I'm uable to ping the server, and tracroute is unable to produce >> a route to it. However, I can telnet to the rsync port on that host >> and it connects.] >> >> When I see that server has been selected, I usually just hit Ctrl-C >> and try again. >> >> Is there any way to configure portage to not use that server? > > Is it possible to add it to your hosts file and point it to local IP? No. Because the name is rsync://rsync.us.gentoo.org/gentoo-portage. > Obviously, if it is a numbered IP then this likely won't work. I could, however, set up a static route for the IP in question and point it to something that isn't listening on the rsync port. That won't make it avoid trying to use that server, but it would make it fail immediately rather than let it crawl along along until it hangs or I hit Ctrl-C. :) -- Grant
Re: [gentoo-user] Blacklist one of the pool's rsync server?
Grant Edwards wrote: > Is there any way for me to "blacklist" a pool rsync server so that > emerge --sync won't try to use it? > > I'm using the sync-url rsync://rsync.us.gentoo.org/gentoo-portage, but > one of the pool's servers is barely usable for me. I don't know if > it's a server problem or if traffic between my and that server is > routed through a Mars orbiter. When any other server is chosen, the > file list scrolls by faster than you can read it. Via the Mars > orbibiter the list ticks by about one file every 5-10 seconds. It > will usually finish (eventually), but sometimes hangs and times out. > It's been like for at least a year or two. > > [FWIW, I'm uable to ping the server, and tracroute is unable to produce > a route to it. However, I can telnet to the rsync port on that host > and it connects.] > > When I see that server has been selected, I usually just hit Ctrl-C > and try again. > > Is there any way to configure portage to not use that server? > Is it possible to add it to your hosts file and point it to local IP? Obviously, if it is a numbered IP then this likely won't work. Dale :-) :-)
[gentoo-user] Blacklist one of the pool's rsync server?
Is there any way for me to "blacklist" a pool rsync server so that emerge --sync won't try to use it? I'm using the sync-url rsync://rsync.us.gentoo.org/gentoo-portage, but one of the pool's servers is barely usable for me. I don't know if it's a server problem or if traffic between my and that server is routed through a Mars orbiter. When any other server is chosen, the file list scrolls by faster than you can read it. Via the Mars orbibiter the list ticks by about one file every 5-10 seconds. It will usually finish (eventually), but sometimes hangs and times out. It's been like for at least a year or two. [FWIW, I'm uable to ping the server, and tracroute is unable to produce a route to it. However, I can telnet to the rsync port on that host and it connects.] When I see that server has been selected, I usually just hit Ctrl-C and try again. Is there any way to configure portage to not use that server? -- Grant
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 02/27/2018 03:30 PM, Neil Bothwick wrote: > > --with-bdeps=n. Once a package is installed, it doesn't matter what > happens to its build deps. FWIW, the rationale for enabling bdeps by default for "upgrade" actions is that the old build deps are already installed; so if there's a major security vulnerability in one of them, you probably want to update it.
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, 27 Feb 2018 18:08:48 +0100, tu...@posteo.de wrote: > > A combination of --changed-deps, --with-bdeps=y and --deep is bound to > > result in plenty of unnecessary re-emerging. > > Hi Neil, > > .andwhat do you suggest instead? --with-bdeps=n. Once a package is installed, it doesn't matter what happens to its build deps. When that package is updated, the build deps will be too if needed, but there's no need to update them several times in the interim. -- Neil Bothwick The road to HAL is paved with good intentions. pgpV8uyCjP397.pgp Description: OpenPGP digital signature
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, Feb 27, 2018 at 1:59 PM, Ian Zimmermanwrote: > On 2018-02-27 12:45, Rich Freeman wrote: > >> I use --with-bdeps=n because I really don't care that much about >> build-time deps, other than stuff that is going to get updated anyway >> like gcc. These packages don't even need to be installed for software >> to work correctly, and if a dev does miss an RDEPEND there is a decent >> chance they'll miss it in DEPEND also. > > Ok, so this looks like the relevant one in the current mess, at least. > But: if you omit --with-bdeps, what happens in the (relatively frequent) > situation when an already installed package is being updated and > acquired a new build dependency, which is not yet installed? > Portage will automatically install the new build dependency first. Anytime you explicitly install a package portage checks all dependencies and ensures they are met. If either a build-time or run-time dependency has changed and is no longer satisfied then portage will take care of that before installing the package. If a build-time dependency changes then by default portage will ignore this unless the package actually needs to be reinstalled. In theory you should never need --with-bdeps=y to actually keep software running. If things break without that option it is likely due to missing dependencies somewhere. That isn't to say that there aren't potential benefits to using --with-bdeps=y and --changed-deps. A package might not NEED its build system updated to run, but if that build system provides some kind of improved optimizations then updating the build system and rebuilding the package could provide runtime benefits. It just shouldn't be needed to actually make the package run. In practice significant optimization improvements aren't that common from regular version upgrades/etc. -- Rich
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
Ian Zimmerman wrote: > On 2018-02-27 11:02, Neil Bothwick wrote: > >> A combination of --changed-deps, --with-bdeps=y and --deep is bound to >> result in plenty of unnecessary re-emerging. > So, what _is_ the recommended set of emerge flags for regular daily or > weekly updates (assuming no binary packages)? > Over the years, I've ended up with this command and options set that seem to work pretty well for most everything. This is in make.conf: EMERGE_DEFAULT_OPTS="--with-bdeps y --backtrack=100 --keep-going -v -j5 --quiet-build=n -1 --unordered-display" After I run eix-sync, I then run emerge -uaDN world and give the updates a looking over, to make sure the USE flags are like I want etc etc. I started out with fewer options but as issues popped up or options were added that made things work better, they were added. Some are done in make.conf to make sure they are the default for every command, unless I override it on the command line. So far, it has resulted in a fairly stable system even if I have some arch packages installed, KDE for example. As usual, you may need something different but that has worked for me and could be a starting point at least. Hope that helps. Dale :-) :-) P. S. On the rare occasion I want to add something to the world file, I either do it directly or use --select y to override the -1 in make.conf. That helps keep the world file from getting cluttered up to no end with things that shouldn't be there.
[gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 2018-02-27 12:45, Rich Freeman wrote: > I use --with-bdeps=n because I really don't care that much about > build-time deps, other than stuff that is going to get updated anyway > like gcc. These packages don't even need to be installed for software > to work correctly, and if a dev does miss an RDEPEND there is a decent > chance they'll miss it in DEPEND also. Ok, so this looks like the relevant one in the current mess, at least. But: if you omit --with-bdeps, what happens in the (relatively frequent) situation when an already installed package is being updated and acquired a new build dependency, which is not yet installed? -- Please don't Cc: me privately on mailing lists and Usenet, if you also post the followup to the list or newsgroup. To reply privately _only_ on Usenet and on broken lists which rewrite From, fetch the TXT record for no-use.mooo.com.
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, Feb 27, 2018 at 1:12 PM, Wols Listswrote: > > If I emerge a new utility program (such as lame), I will change my > global flags to tell other programs to use it. That is what > --changed-deps is for - so the programs that were originally compiled > without support will be re-compiled with support. No ebuilds have been > changed anywhere ... > You're thinking of --changed-use. -- Rich
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 27/02/18 17:43, Michael Orlitzky wrote: > The --changed-deps flag, on the other hand, is a crutch for when > developers make in-place edits to ebuilds and don't make the necessary > revision bump. I believe the --changed-deps flag is ALSO for USERS who want to change settings on their computer. If I emerge a new utility program (such as lame), I will change my global flags to tell other programs to use it. That is what --changed-deps is for - so the programs that were originally compiled without support will be re-compiled with support. No ebuilds have been changed anywhere ... Cheers, Wol
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, Feb 27, 2018 at 12:43 PM, Michael Orlitzkywrote: > On 02/27/2018 12:05 PM, Ian Zimmerman wrote: >> >> When I read this, I realize I don't understand the difference between >> these two options. Or to be more accurate, I know that --deep means >> looking at dependencies beyond the first level; but isn't that just a >> superset of those found by --changed-deps? > > You're right about --deep. > > The --changed-deps flag, on the other hand, is a crutch for when > developers make in-place edits to ebuilds and don't make the necessary > revision bump. The revision bump is required, but sometimes people > forget or are just assholes and don't care that this is a huge hassle > for end users. When you use --changed-deps, portage scans the tree for > those sorts of mistakes, and rebuilds the packages that have been > changed in-place, without regard to where in the dependency graph they lie. > To be fair --changed-deps also forces builds that aren't really necessary. Not all dep changes necessarily require a revbump. However, you are right that these do get missed when they are necessary, and really devs should just default to revbumping except when they know it isn't needed. I'm willing to bet that 95% of the --changed-deps rebuilds aren't necessary. The problem is that when one is necessary it tends to result in problems that the average user is going to struggle to troubleshoot. I suspect most users would rather do a few more rebuilds than to have stuff break mysteriously and be forced to try to figure out what combination of package rebuilds is necessary to get it working, and most people probably would prefer extra builds to breakage even if it was easy to troubleshoot. -- Rich
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, Feb 27, 2018 at 12:05 PM, Ian Zimmermanwrote: > On 2018-02-27 16:44, Paul Colquhoun wrote: > >> > > Yes, I use --deep. I've run into cases many times in the past >> > > where portage was skipping updates unless I used --deep. >> > >> > You might want to avoid combining both --deep and --changed-deps. > > When I read this, I realize I don't understand the difference between > these two options. Or to be more accurate, I know that --deep means > looking at dependencies beyond the first level; but isn't that just a > superset of those found by --changed-deps? If so, how would two runs of > emerge (each with one of these options) help? Based on my naive > hypothesis above, emerge --deep would do everything that emerge --deep > --changed-deps would. > Yeah, I mixed up --deep and --with-bdeps=y. I would use --deep across the board - you want your runtime deps to be up to date. In my regular update script I have: ionice -c 3 nice -n 15 emerge -auDkv --changed-use --keep-going --with-bdeps=n --changed-deps --binpkg-changed-deps=y --backtrack=100 world I'll note I haven't had a particularly exciting number of package rebuilds in the last few days. --binpkg-changed-deps is probably not essential as this setting is actually the default, but since I use binary packages as much as I can (built with cron the night before) I don't want old stale packages to be used if I'm using --changed-deps, since that is just a waste of disk IO. I use --changed-deps because I don't quite trust devs to be revbumping stuff when they should. Many will argue it shouldn't be necessary, and if we were better with QA I'd agree. I use --with-bdeps=n because I really don't care that much about build-time deps, other than stuff that is going to get updated anyway like gcc. These packages don't even need to be installed for software to work correctly, and if a dev does miss an RDEPEND there is a decent chance they'll miss it in DEPEND also. Once in a blue moon I'll run into a situation where something has a bad build dependency or a build system otherwise gets broken and I'll go ahead and do a one-time update with --with-bdeps=y. I probably would leave out --changed-deps in this case unless things are still broken. Though, I do this so rarely I suspect it wouldn't make much difference as so many build-time deps are going to get updated anyway. -- Rich
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 02/27/2018 12:05 PM, Ian Zimmerman wrote: > > When I read this, I realize I don't understand the difference between > these two options. Or to be more accurate, I know that --deep means > looking at dependencies beyond the first level; but isn't that just a > superset of those found by --changed-deps? You're right about --deep. The --changed-deps flag, on the other hand, is a crutch for when developers make in-place edits to ebuilds and don't make the necessary revision bump. The revision bump is required, but sometimes people forget or are just assholes and don't care that this is a huge hassle for end users. When you use --changed-deps, portage scans the tree for those sorts of mistakes, and rebuilds the packages that have been changed in-place, without regard to where in the dependency graph they lie.
[gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 2018-02-27 11:02, Neil Bothwick wrote: > A combination of --changed-deps, --with-bdeps=y and --deep is bound to > result in plenty of unnecessary re-emerging. So, what _is_ the recommended set of emerge flags for regular daily or weekly updates (assuming no binary packages)? -- Please don't Cc: me privately on mailing lists and Usenet, if you also post the followup to the list or newsgroup. To reply privately _only_ on Usenet and on broken lists which rewrite From, fetch the TXT record for no-use.mooo.com.
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 02/27 11:02, Neil Bothwick wrote: > On Tue, 27 Feb 2018 12:15:24 +0200, Nikos Chantziaras wrote: > > > >> luky you...I got 462 packages to recompile... > > > > > > Ooh! Bloat warning! > > > > It got even worse just today. > > > > Arch Linux is starting to look really, really tasty right about now... > > A combination of --changed-deps, --with-bdeps=y and --deep is bound to > result in plenty of unnecessary re-emerging. > > > -- > Neil Bothwick > > For every action, there is an equal and opposite malfunction. Hi Neil, .andwhat do you suggest instead? Cheers Meino
[gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 2018-02-27 16:44, Paul Colquhoun wrote: > > > Yes, I use --deep. I've run into cases many times in the past > > > where portage was skipping updates unless I used --deep. > > > > You might want to avoid combining both --deep and --changed-deps. When I read this, I realize I don't understand the difference between these two options. Or to be more accurate, I know that --deep means looking at dependencies beyond the first level; but isn't that just a superset of those found by --changed-deps? If so, how would two runs of emerge (each with one of these options) help? Based on my naive hypothesis above, emerge --deep would do everything that emerge --deep --changed-deps would. Or not? Please explain. -- Please don't Cc: me privately on mailing lists and Usenet, if you also post the followup to the list or newsgroup. To reply privately _only_ on Usenet and on broken lists which rewrite From, fetch the TXT record for no-use.mooo.com.
Re: [gentoo-user] emerge @world and package.mask ?
Am 27.02.2018 um 14:29 schrieb Helmut Jarausch: > I don't understand portage (any more). > > I want to keep dev-qt/qt-meta-4.8.6 (QT4) which requires > > media-libs/phonon[qt4] > > I do have media-libs/phonon-4.9.1-r1[qt4,qt5] installed here. > And in /etc/portage/package.mask I have >> media-libs/phonon-4.9.9 >> media-libs/phonon-vlc-4.9.9> > But still, > > nice -19 emerge -v1 -j16 --update --keep-going --tree --changed-use > --unordered-display --verbose-conflicts --deep --with-bdeps?y @world > > requires me to remove that masks since it's going to upgrade > media-libs/phonon and media-libs/phonon-vlc. > Why doesn't emerge respect me my masks? Hi, maybe you want something like this in your package.mask if you don't want phonon version > 4.10: >=media-libs/phonon-4.10 >=media-libs/phonon-vlc-0.10 Also there's no phonon-vlc-4.9.9. phonon-vlc only goes up to 0.10.1, that's why portage tries to update phonon-vlc. Steven
[gentoo-user] emerge @world and package.mask ?
Hi, I don't understand portage (any more). I want to keep dev-qt/qt-meta-4.8.6 (QT4) which requires media-libs/phonon[qt4] I do have media-libs/phonon-4.9.1-r1[qt4,qt5] installed here. And in /etc/portage/package.mask I have media-libs/phonon-4.9.9 media-libs/phonon-vlc-4.9.9 But still, nice -19 emerge -v1 -j16 --update --keep-going --tree --changed-use --unordered-display --verbose-conflicts --deep --with-bdeps?y @world requires me to remove that masks since it's going to upgrade media-libs/phonon and media-libs/phonon-vlc. Why doesn't emerge respect me my masks? Many thanks for some hints, Helmut Output of emerge ... @world media-libs/phonon:0 (media-libs/phonon-4.10.0:0/0::gentoo, ebuild scheduled for merge) pulled in by >?media-libs/phonon-4.10.0 required by (media-libs/phonon-vlc-0.10.1:0/0::gentoo, ebuild scheduled for merge) ^^ ^^ (media-libs/phonon-4.9.1-r1:0/0::gentoo, installed) pulled in by media-libs/phonon[qt4] required by (dev-qt/qt-meta-4.8.6:4/4::gentoo, installed) ^^^ >?media-libs/phonon-4.9.0[qt4?,qt5?] required by (media-libs/phonon-vlc-0.9.1-r1:0/0::gentoo, ebuild scheduled for merge) media-libs/phonon-vlc:0 (media-libs/phonon-vlc-0.10.1:0/0::gentoo, ebuild scheduled for merge) pulled in by (no parents that aren't satisfied by other packages in this slot) (media-libs/phonon-vlc-0.9.1-r1:0/0::gentoo, ebuild scheduled for merge) pulled in by >?media-libs/phonon-vlc-0.9.0[qt4?,qt5?] required by (media-libs/phonon-4.9.1-r1:0/0::gentoo, installed)=
Re: [gentoo-user] Re: Is --changed-deps going to be *that* useless?
On Tue, 27 Feb 2018 12:15:24 +0200, Nikos Chantziaras wrote: > >> luky you...I got 462 packages to recompile... > > > > Ooh! Bloat warning! > > It got even worse just today. > > Arch Linux is starting to look really, really tasty right about now... A combination of --changed-deps, --with-bdeps=y and --deep is bound to result in plenty of unnecessary re-emerging. -- Neil Bothwick For every action, there is an equal and opposite malfunction. pgphUAv8JlJmC.pgp Description: OpenPGP digital signature
[gentoo-user] Re: Is --changed-deps going to be *that* useless?
On 27/02/18 01:55, Peter Humphrey wrote: On Monday, 26 February 2018 18:42:33 GMT tu...@posteo.de wrote: Hi Peter, luky you...I got 462 packages to recompile... Ooh! Bloat warning! It got even worse just today. Arch Linux is starting to look really, really tasty right about now...