Re: when startup delays become bugs
Adam Williamson (awill...@redhat.com) said: Isn't anaconda-tools the group used to ensure that things anaconda may need to install for particular hardware are present on images? Live DVD/install images, yes. Bill -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On 05/18/2013 05:17 AM, Nicolas Mailhot wrote: Le Sam 18 mai 2013 05:39, T.C. Hollingsworth a écrit : On Fri, May 17, 2013 at 7:34 PM, Adam Williamson awill...@redhat.com wrote: You could transfer the install to a system which contains a dmraid array, or add a dmraid array to an existing install (I think this thread has been considering only the case of the installed system itself being on the RAID array). Of course, the case where the installed system is not on the RAID array is much less 'urgent' - your system doesn't stop working, you just have to figure out you need to enable the service / install the tool in order to see the array. Yeah, but you'd need to do some manual configuration there anyway. Adding `yum install dmraid` to that isn't really a massive burden. That's really a terrible argument. It's why you get so many one thousand paper cut moments in IT nowadays. IT failures tend to be either because the design is too simplistic or too complex. Windows is in the first category: it only works in the specific hardware configuration it was originally installed---although I suspect that it's a deliberate choice to prevent illegal mass cloning, rather than inability to make it more portable. Of course Linux has built-in drivers for most popular chipsets so it tends to boot on any hardware you put it on, at least in a basic mode. I am quite happy with that---it allows me to do what I need to, in an incremental way, rather than telling me that I am SOL and should have started with a 'sysprep' but it's too late now. I would say let's count our blessings and not expect that every possible combination of hardware components will be supported in an optimal way. Just as long as the whole thing reliably comes up, I'd be happy. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Am 17.05.2013 14:45, schrieb Lennart Poettering: Here we go again: https://www.dropbox.com/s/gdrdvq0kovucpsp/bootchart-20130517-1530.svg There is something really fishy... For example, from 8s to 10s the machine goes pretty much entirely idle... As if there was something doing sleep(2) or so... maybe something related with the bugreport i wrote a few minutes ago where systemd-analyze blame does not display services with type=simple https://bugzilla.redhat.com/show_bug.cgi?id=964760 signature.asc Description: OpenPGP digital signature -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
Le Sam 18 mai 2013 05:39, T.C. Hollingsworth a écrit : On Fri, May 17, 2013 at 7:34 PM, Adam Williamson awill...@redhat.com wrote: You could transfer the install to a system which contains a dmraid array, or add a dmraid array to an existing install (I think this thread has been considering only the case of the installed system itself being on the RAID array). Of course, the case where the installed system is not on the RAID array is much less 'urgent' - your system doesn't stop working, you just have to figure out you need to enable the service / install the tool in order to see the array. Yeah, but you'd need to do some manual configuration there anyway. Adding `yum install dmraid` to that isn't really a massive burden. That's really a terrible argument. It's why you get so many one thousand paper cut moments in IT nowadays. -- Nicolas Mailhot -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
Am 18.05.2013 11:17, schrieb Nicolas Mailhot: Le Sam 18 mai 2013 05:39, T.C. Hollingsworth a écrit : On Fri, May 17, 2013 at 7:34 PM, Adam Williamson awill...@redhat.com wrote: You could transfer the install to a system which contains a dmraid array, or add a dmraid array to an existing install (I think this thread has been considering only the case of the installed system itself being on the RAID array). Of course, the case where the installed system is not on the RAID array is much less 'urgent' - your system doesn't stop working, you just have to figure out you need to enable the service / install the tool in order to see the array. Yeah, but you'd need to do some manual configuration there anyway. Adding `yum install dmraid` to that isn't really a massive burden. That's really a terrible argument. It's why you get so many one thousand paper cut moments in IT nowadays to it is a *very good argument* nobody needs *everything and all* in a *core setup* and i get my paper cut moments where people trying to make defaults and core setups idiot proof which will *never* success signature.asc Description: OpenPGP digital signature -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
Hi, On 05/17/2013 01:21 AM, Lennart Poettering wrote: On Thu, 16.05.13 13:44, Adam Williamson (awill...@redhat.com) wrote: On Thu, 2013-05-16 at 20:41 +, Jóhann B. Guðmundsson wrote: On 05/16/2013 08:17 PM, Bill Nottingham wrote: We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? dmraid is hardly enterprise, though. It's used for all firmware RAID implementations aside from Intel's, and it's not unusual for any random system to be using firmware RAID. As I understood this non-intel fakeraid is actually pretty much the exception... I'm afraid that is not the case, back when I worked on anaconda we had tons and tons of dmraid (so non intel firmware raid) related bugs, mostly in dmraid itself (*), and yes this is 2 years ago but I don't believe the situation will have changed drastically all amd motherboards for example will still need dmraid if they are using the motherboard build in raid, as well as jmicron controllers slapped on for extra sata ports, and even some pci(-express) add in raid cards need dmraid because they are just firmware raid. Also various raid + ssd all in one contraptions are being build, which likely also need dmraid. So at least in the live images having dmraid by default is a must IMHO, otherwise we may end up writing to one disk of a mirror without marking the mirror dirty, which is not good, not good at all. But I think we can still do better here while keeping dmraid by default, it is highly highly unlikely for an installed system which is not using dmraid to all of a sudden get a dmraid set without a re-install. It is possible, but this requires an advanced user to be doing all kind of manual tweaking, and we can just document that in this case one more tweak is necessary. So I would like to suggest that we simple make the dmraid service write a (very simple) config file at the end of its first run (so its writes the config file only if it does not exist), and then use the contents of that file to conditionally start dmraid from then on. Note I believe this really needs to be a config file, and not a if no dmraid-sets where found disable service kind of magic, so that if a dmraid set was available but temporarily becomes unavailable, we don't end up disabling the service and then if the set is fixed / restored, we still don't do the right thing. If others (esp Heinz and Peter, added to the CC) agree this would be a good solution to no longer dragging in systemd-udev-settle into every systems boot, I think we should schedule fixing this for F-20. Regards, Hans *) Not dmraid's fault, just the price we pay for reverse-engineering these firmware raid on disk meta data format. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Hi, On 05/17/2013 01:31 AM, Lennart Poettering wrote: On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? Yes, it already drags in for example iscsi-initiator-utils automatically when iscsi disks are used during install. IIRC doing the same for dmraid should not be hard. Also see my comment in another thread, where I offer a solution which would work with the always popular livecd installs too, and thus is better IMHO. Regards, Hans -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 17.05.13 07:09, Heiko Adams (heiko.ad...@gmail.com) wrote: Am 17.05.2013 01:09, schrieb Lennart Poettering: For the super slow run above I'd be quite interested to have a look at the bootchart actually. (Heiko? Can you upload that?) Lennart https://www.dropbox.com/s/95dzgklajrgaz4n/bootchart-20130517-0756.svg BTW: Is it correct that /dev/mapper/VolGroup-lv_home can't be found with systemd-bootchart? Hmm, for some reason the whole thing is truncated... There are intervals of 2s where the whole machine doesn't do anything apparently... Can you rerun this with initcall_debug? Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On Fri, 17.05.13 09:05, Hans de Goede (hdego...@redhat.com) wrote: Hi, On 05/17/2013 01:21 AM, Lennart Poettering wrote: On Thu, 16.05.13 13:44, Adam Williamson (awill...@redhat.com) wrote: On Thu, 2013-05-16 at 20:41 +, Jóhann B. Guðmundsson wrote: On 05/16/2013 08:17 PM, Bill Nottingham wrote: We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? dmraid is hardly enterprise, though. It's used for all firmware RAID implementations aside from Intel's, and it's not unusual for any random system to be using firmware RAID. As I understood this non-intel fakeraid is actually pretty much the exception... I'm afraid that is not the case, back when I worked on anaconda we had tons and tons of dmraid (so non intel firmware raid) related bugs, mostly in dmraid itself (*), and yes this is 2 years ago but I don't believe the situation will have changed drastically all amd motherboards for example will still need dmraid if they are using the motherboard build in raid, as well as jmicron controllers slapped on for extra sata ports, and even some pci(-express) add in raid cards need dmraid because they are just firmware raid. Also various raid + ssd all in one contraptions are being build, which likely also need dmraid. So at least in the live images having dmraid by default is a must IMHO, otherwise we may end up writing to one disk of a mirror without marking the mirror dirty, which is not good, not good at all. But I think we can still do better here while keeping dmraid by default, it is highly highly unlikely for an installed system which is not using dmraid to all of a sudden get a dmraid set without a re-install. It is possible, but this requires an advanced user to be doing all kind of manual tweaking, and we can just document that in this case one more tweak is necessary. So I would like to suggest that we simple make the dmraid service write a (very simple) config file at the end of its first run (so its writes the config file only if it does not exist), and then use the contents of that file to conditionally start dmraid from then on. Note I believe this really needs to be a config file, and not a if no dmraid-sets where found disable service kind of magic, so that if a dmraid set was available but temporarily becomes unavailable, we don't end up disabling the service and then if the set is fixed / restored, we still don't do the right thing. If others (esp Heinz and Peter, added to the CC) agree this would be a good solution to no longer dragging in systemd-udev-settle into every systems boot, I think we should schedule fixing this for F-20. I have filed a bug for this now: https://bugzilla.redhat.com/show_bug.cgi?id=964172 I also filed this bug against anaconda, so that for the non-livecd installs we don't even get dmraid installed... https://bugzilla.redhat.com/show_bug.cgi?id=964175 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Am 17.05.2013 14:18, schrieb Lennart Poettering: On Fri, 17.05.13 07:09, Heiko Adams (heiko.ad...@gmail.com) wrote: Am 17.05.2013 01:09, schrieb Lennart Poettering: For the super slow run above I'd be quite interested to have a look at the bootchart actually. (Heiko? Can you upload that?) Lennart https://www.dropbox.com/s/95dzgklajrgaz4n/bootchart-20130517-0756.svg BTW: Is it correct that /dev/mapper/VolGroup-lv_home can't be found with systemd-bootchart? Hmm, for some reason the whole thing is truncated... There are intervals of 2s where the whole machine doesn't do anything apparently... Can you rerun this with initcall_debug? Lennart Here we go again: https://www.dropbox.com/s/gdrdvq0kovucpsp/bootchart-20130517-1530.svg -- Regards, Heiko Adams -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 17.05.13 14:38, Heiko Adams (heiko.ad...@gmail.com) wrote: Am 17.05.2013 14:18, schrieb Lennart Poettering: On Fri, 17.05.13 07:09, Heiko Adams (heiko.ad...@gmail.com) wrote: Am 17.05.2013 01:09, schrieb Lennart Poettering: For the super slow run above I'd be quite interested to have a look at the bootchart actually. (Heiko? Can you upload that?) Lennart https://www.dropbox.com/s/95dzgklajrgaz4n/bootchart-20130517-0756.svg BTW: Is it correct that /dev/mapper/VolGroup-lv_home can't be found with systemd-bootchart? Hmm, for some reason the whole thing is truncated... There are intervals of 2s where the whole machine doesn't do anything apparently... Can you rerun this with initcall_debug? Lennart Here we go again: https://www.dropbox.com/s/gdrdvq0kovucpsp/bootchart-20130517-1530.svg There is something really fishy... For example, from 8s to 10s the machine goes pretty much entirely idle... As if there was something doing sleep(2) or so... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
Once upon a time, Lennart Poettering mzerq...@0pointer.de said: I also filed this bug against anaconda, so that for the non-livecd installs we don't even get dmraid installed... I know there's always a goal of shrinking the base install, but I'm afraid we're going overboard. One nice thing about Linux vs. Windows has always been how easy it is to move an install from one physical system to another (install in one and move the hard drive, system dies but drive is okay so you replace mboard, etc.). By cutting out all the hardware support except for what is in the system at install, it becomes more difficult (like Windows) to deal with any problems, hardware upgrades, etc. down the road. -- Chris Adams li...@cmadams.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On Fri, May 17, 2013 at 08:43:36AM -0500, Chris Adams wrote: By cutting out all the hardware support except for what is in the system at install, it becomes more difficult (like Windows) to deal with any problems, hardware upgrades, etc. down the road. Agreed, but I think hardware-which-requires-weird-daemons is an okay place to draw the line for @core, at least. Maybe it *should* be in @standard? -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On Fri, May 17, 2013 at 08:43:36 -0500, Chris Adams li...@cmadams.net wrote: By cutting out all the hardware support except for what is in the system at install, it becomes more difficult (like Windows) to deal with any problems, hardware upgrades, etc. down the road. Note that the change to defaulting to hostonly for dracut has already sent us down that path. People who want to be able to move disks without needing to use rescue mode to fix things up on the new hardware, will already need to be doing some work. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 2013-05-17 at 09:07 +0200, Hans de Goede wrote: Hi, On 05/17/2013 01:31 AM, Lennart Poettering wrote: On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? Yes, it already drags in for example iscsi-initiator-utils automatically when iscsi disks are used during install. IIRC doing the same for dmraid should not be hard. This has been in place since somewhere around F11 or F12. You just can't tell since dmraid is in base. There is still the anaconda-tools group, which I forget the purpose of (live media?). -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 17.05.13 10:18, David Lehman (dleh...@redhat.com) wrote: On Fri, 2013-05-17 at 09:07 +0200, Hans de Goede wrote: Hi, On 05/17/2013 01:31 AM, Lennart Poettering wrote: On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? Yes, it already drags in for example iscsi-initiator-utils automatically when iscsi disks are used during install. IIRC doing the same for dmraid should not be hard. This has been in place since somewhere around F11 or F12. You just can't tell since dmraid is in base. There is still the anaconda-tools group, which I forget the purpose of (live media?). So, are you saying we could simply drop dmraid from base, and anaconda would do the right thing and install it when dmraid is used for the installation? Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 2013-05-17 at 17:45 +0200, Lennart Poettering wrote: On Fri, 17.05.13 10:18, David Lehman (dleh...@redhat.com) wrote: On Fri, 2013-05-17 at 09:07 +0200, Hans de Goede wrote: Hi, On 05/17/2013 01:31 AM, Lennart Poettering wrote: On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? Yes, it already drags in for example iscsi-initiator-utils automatically when iscsi disks are used during install. IIRC doing the same for dmraid should not be hard. This has been in place since somewhere around F11 or F12. You just can't tell since dmraid is in base. There is still the anaconda-tools group, which I forget the purpose of (live media?). So, are you saying we could simply drop dmraid from base, and anaconda would do the right thing and install it when dmraid is used for the installation? Yes. Likewise lvm2, mdadm, cryptsetup, e2fsprogs, and whatever other storage-specific packages are in there. Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 16.05.2013 21:29, Heiko Adams wrote: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service And I don't have a clue why the hell I've got an avahi-daemon.service running on every startup an why it take so extreme long to start. … Also your 'lightdm' seems to trying to compete with 'gdm', I mean really what are you two doing!? :) systemd-analyze blame | grep lightdm 467ms lightdm.service poma -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 17.05.2013 14:45, Lennart Poettering wrote: There is something really fishy... … Yeah! The week *end* has just begun! :) poma -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 2013-05-17 at 10:18 -0500, David Lehman wrote: On Fri, 2013-05-17 at 09:07 +0200, Hans de Goede wrote: Hi, On 05/17/2013 01:31 AM, Lennart Poettering wrote: On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? Yes, it already drags in for example iscsi-initiator-utils automatically when iscsi disks are used during install. IIRC doing the same for dmraid should not be hard. This has been in place since somewhere around F11 or F12. You just can't tell since dmraid is in base. There is still the anaconda-tools group, which I forget the purpose of (live media?). Isn't anaconda-tools the group used to ensure that things anaconda may need to install for particular hardware are present on images? -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/16/2013 02:39 PM, Lennart Poettering wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service Well, but only fsck itself knows that and can determine this from the superblock. Hence we have to start it first and it will then exit quickly if the fs wasn't dirty. Note that these times might be misleading: if fsck takes this long to check the superblock and exit this might be a result of something else which runs in parallel monopolizing CPU or IO (for example readahead), and might not actually be fsck's own fault. We really should not need to run fsck on boot unless the mount fails. Might save some time at the cost of a bit of extra complexity? Ric and no oops, so this seems unnecessary: 1.092s abrt-uefioops.service https://bugzilla.redhat.com/show_bug.cgi?id=963182 and I'm not using LVM so these seem unnecessary: 2.783s lvm2-monitor.service 489ms systemd-udev-settle.service 15ms lvm2-lvmetad.service How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? $ repoquery --qf=%{sourcerpm} --whatprovides '*/lib/systemd/system/lvm2-monitor.service' lvm2-2.02.98-8.fc19.src.rpm Please file a bug against the lvm2 package. And make sure to add it to: https://bugzilla.redhat.com/show_bug.cgi?id=963210 Hmm, on your machine, what does systemctl show -p WantedBy -p RequiredBy systemd-udev-settle.service show? This will tell us which package is actually responsible for pulling in systemd-udev-settle.service. Thanks! Lennart -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 05:09 PM, Rahul Sundaram wrote: On 05/15/2013 04:39 PM, Przemek Klosowski wrote: I was planning to upgrade to F19 soon, and I kind-of care about the data on that system (I have backup, but corruption would not be welcome, just for the lost time reason). Do people recommend sticking with it? holding off the upgrade? switching back to ext4? https://btrfs.wiki.kernel.org/index.php/FAQ#Is_btrfs_stable.3F is a reasonably good answer to that question. I would note that traditional linux filesystems have a very long history and time to mature including XFS and Ext4 and only check for metadata integrity unlike Btrfs. If you have a good (and frequent) backup system, it might be worth riding it out but you will have to evaluate the risks and make that determination of your own. FWIW, I am using Btrfs without any (known) issues for a while but I also have almost no data in my laptop that I cannot afford to lose since everything important is an external backup disk and really important files are in another laptop as well. Rahul The right answer is of course to always back up every file system since disk dies, bugs eat data, etc :) If you have backups and want to help btrfs settle in, it is really valuable to have people on it. btrfs is getting more stable, but for critical data I would suggest using xfs or ext4. Ric -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 17, 2013, at 2:38 PM, Ric Wheeler rwhee...@redhat.com wrote: On 05/16/2013 02:39 PM, Lennart Poettering wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service Well, but only fsck itself knows that and can determine this from the superblock. Hence we have to start it first and it will then exit quickly if the fs wasn't dirty. Note that these times might be misleading: if fsck takes this long to check the superblock and exit this might be a result of something else which runs in parallel monopolizing CPU or IO (for example readahead), and might not actually be fsck's own fault. We really should not need to run fsck on boot unless the mount fails. Might save some time at the cost of a bit of extra complexity? Seems some extra complexity is needed anyway since the way to deal with file system problems differs with the various fs's. XFS and Btrfs fsck's are noops. XFS needs xfs_repair run, and Btrfs maybe needs to be remounted with -o degraded, depending on the nature of the mount failure since most problems are autorecovered from during mount. Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 5/17/13 3:38 PM, Ric Wheeler wrote: On 05/16/2013 02:39 PM, Lennart Poettering wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service Well, but only fsck itself knows that and can determine this from the superblock. Hence we have to start it first and it will then exit quickly if the fs wasn't dirty. Note that these times might be misleading: if fsck takes this long to check the superblock and exit this might be a result of something else which runs in parallel monopolizing CPU or IO (for example readahead), and might not actually be fsck's own fault. We really should not need to run fsck on boot unless the mount fails. Might save some time at the cost of a bit of extra complexity? well, ext[34]a are special little snowflakes. ;) Since forever, we've called fsck on boot, and e2fsck replays the log in userspace if needed. If the fs isn't marked as having encountered an error during the previous mount (or, historically, having had too many mounts or too much time since last fsck), then nothing else happens. It shouldn't take a whole ton of time, but could, depending on whether the log was dirty, then the size of the log and speed of the disk I suppose. When it took 4s above, was that a from a clean reboot (i.e. was the journal dirty?) -Eric Ric and no oops, so this seems unnecessary: 1.092s abrt-uefioops.service https://bugzilla.redhat.com/show_bug.cgi?id=963182 and I'm not using LVM so these seem unnecessary: 2.783s lvm2-monitor.service 489ms systemd-udev-settle.service 15ms lvm2-lvmetad.service How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? $ repoquery --qf=%{sourcerpm} --whatprovides '*/lib/systemd/system/lvm2-monitor.service' lvm2-2.02.98-8.fc19.src.rpm Please file a bug against the lvm2 package. And make sure to add it to: https://bugzilla.redhat.com/show_bug.cgi?id=963210 Hmm, on your machine, what does systemctl show -p WantedBy -p RequiredBy systemd-udev-settle.service show? This will tell us which package is actually responsible for pulling in systemd-udev-settle.service. Thanks! Lennart -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 5/17/13 3:58 PM, Chris Murphy wrote: On May 17, 2013, at 2:38 PM, Ric Wheeler rwhee...@redhat.com wrote: On 05/16/2013 02:39 PM, Lennart Poettering wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service Well, but only fsck itself knows that and can determine this from the superblock. Hence we have to start it first and it will then exit quickly if the fs wasn't dirty. Note that these times might be misleading: if fsck takes this long to check the superblock and exit this might be a result of something else which runs in parallel monopolizing CPU or IO (for example readahead), and might not actually be fsck's own fault. We really should not need to run fsck on boot unless the mount fails. Might save some time at the cost of a bit of extra complexity? Seems some extra complexity is needed anyway since the way to deal with file system problems differs with the various fs's. XFS and Btrfs fsck's are noops. XFS needs xfs_repair run, and Btrfs maybe needs to be remounted with -o degraded, depending on the nature of the mount failure since most problems are autorecovered from during mount. fsck.xfs is a no-op because of the xfs approach that it's a journaling filesystem, so the mount-time recovery is simply replay the log you're good. If you have a corrupt filesystem (as opposed to a not-cleanly-unmounted filesystem), xfs_repair is an administrative action, not a boot-time auto-initiated initscript action. -Eric Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 17, 2013, at 3:56 PM, Eric Sandeen sand...@redhat.com wrote: On 5/17/13 3:58 PM, Chris Murphy wrote: Seems some extra complexity is needed anyway since the way to deal with file system problems differs with the various fs's. XFS and Btrfs fsck's are noops. XFS needs xfs_repair run, and Btrfs maybe needs to be remounted with -o degraded, depending on the nature of the mount failure since most problems are autorecovered from during mount. fsck.xfs is a no-op because of the xfs approach that it's a journaling filesystem, so the mount-time recovery is simply replay the log you're good. If you have a corrupt filesystem (as opposed to a not-cleanly-unmounted filesystem), xfs_repair is an administrative action, not a boot-time auto-initiated initscript action. So if the boot fails due to reasons other than an unclean mount, with xfs the user needs a rescue environment of some sort. At the moment, it's similar with Btrfs in that what to do next depends on the problem. Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 17, 2013, at 3:53 PM, Eric Sandeen sand...@redhat.com wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service When it took 4s above, was that a from a clean reboot (i.e. was the journal dirty?) Clean. And it's a new file system, created within the hour of the time test. I also don't understand why there are two instances. There's only one ext4 file system on the computer. Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 5/17/13 5:29 PM, Chris Murphy wrote: On May 17, 2013, at 3:56 PM, Eric Sandeen sand...@redhat.com wrote: On 5/17/13 3:58 PM, Chris Murphy wrote: Seems some extra complexity is needed anyway since the way to deal with file system problems differs with the various fs's. XFS and Btrfs fsck's are noops. XFS needs xfs_repair run, and Btrfs maybe needs to be remounted with -o degraded, depending on the nature of the mount failure since most problems are autorecovered from during mount. fsck.xfs is a no-op because of the xfs approach that it's a journaling filesystem, so the mount-time recovery is simply replay the log you're good. If you have a corrupt filesystem (as opposed to a not-cleanly-unmounted filesystem), xfs_repair is an administrative action, not a boot-time auto-initiated initscript action. So if the boot fails due to reasons other than an unclean mount, unclean mount doesn't fail boot ;) with xfs the user needs a rescue environment of some sort. At the moment, it's similar with Btrfs in that what to do next depends on the problem. Isn't that always the case? :) Same is true for e2fsck, TBH. Things can go wrong... initramfs contains tools for all these filesystems, AFAIK. -Eric (sorry, I think I'm taking this thread off-topic) Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 5/17/13 5:39 PM, Chris Murphy wrote: On May 17, 2013, at 3:53 PM, Eric Sandeen sand...@redhat.com wrote: On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service When it took 4s above, was that a from a clean reboot (i.e. was the journal dirty?) Clean. And it's a new file system, created within the hour of the time test. I also don't understand why there are two instances. There's only one ext4 file system on the computer. Can't imagine why it takes 4s then, need finer-grained tracing I guess... If I e2fsck a clean fs here it takes about 0.1s. Most of its time is spent in read(); several reads get about 100k of data from the device, and that's about it. As a sanity check I suppose you could try e2fsck from a rescue environment and see if it still takes that long, or of there is other overhead / interaction slowing it down. -Eric Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Le jeudi 16 mai 2013 à 14:44 -0400, john.flor...@dart.biz a écrit : From: Adam Williamson awill...@redhat.com How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? Yes. 'rpm -qf /lib/systemd/system/foo.service'. I'd actually suggest doing: rpm -qif /lib/systemd/system/foo.service ... and noting the source rpm. For the sake of over optimisation : $ rpm --queryformat '%{SOURCERPM}\n' -qf /lib/systemd/system/foo.service foo-1.2-7.fc19.src.rpm -- Michael Scherer -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On May 17, 2013 6:43 AM, Chris Adams li...@cmadams.net wrote: Once upon a time, Lennart Poettering mzerq...@0pointer.de said: I also filed this bug against anaconda, so that for the non-livecd installs we don't even get dmraid installed... I know there's always a goal of shrinking the base install, but I'm afraid we're going overboard. One nice thing about Linux vs. Windows has always been how easy it is to move an install from one physical system to another (install in one and move the hard drive, system dies but drive is okay so you replace mboard, etc.). By cutting out all the hardware support except for what is in the system at install, it becomes more difficult (like Windows) to deal with any problems, hardware upgrades, etc. down the road. How does this change affect that? If you slap a disk into another machine it's not going to magically start using RAID... -T.C. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On Fri, 2013-05-17 at 19:17 -0700, T.C. Hollingsworth wrote: On May 17, 2013 6:43 AM, Chris Adams li...@cmadams.net wrote: Once upon a time, Lennart Poettering mzerq...@0pointer.de said: I also filed this bug against anaconda, so that for the non-livecd installs we don't even get dmraid installed... I know there's always a goal of shrinking the base install, but I'm afraid we're going overboard. One nice thing about Linux vs. Windows has always been how easy it is to move an install from one physical system to another (install in one and move the hard drive, system dies but drive is okay so you replace mboard, etc.). By cutting out all the hardware support except for what is in the system at install, it becomes more difficult (like Windows) to deal with any problems, hardware upgrades, etc. down the road. How does this change affect that? If you slap a disk into another machine it's not going to magically start using RAID... You could transfer the install to a system which contains a dmraid array, or add a dmraid array to an existing install (I think this thread has been considering only the case of the installed system itself being on the RAID array). Of course, the case where the installed system is not on the RAID array is much less 'urgent' - your system doesn't stop working, you just have to figure out you need to enable the service / install the tool in order to see the array. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs (dmraid)
On Fri, May 17, 2013 at 7:34 PM, Adam Williamson awill...@redhat.com wrote: You could transfer the install to a system which contains a dmraid array, or add a dmraid array to an existing install (I think this thread has been considering only the case of the installed system itself being on the RAID array). Of course, the case where the installed system is not on the RAID array is much less 'urgent' - your system doesn't stop working, you just have to figure out you need to enable the service / install the tool in order to see the array. Yeah, but you'd need to do some manual configuration there anyway. Adding `yum install dmraid` to that isn't really a massive burden. I could have another drive in my new system that uses ZFS, but I don't think we should add zfs-fuse to @core just so any random hard drive with Fedora installed that I happen to want to pop in there can read it off the bat. I love that I can put a drive with Fedora on it into any system and have it boot, but my expectations in this regard end with being able to log in. I do not expect any random new hardware I have in that new system to just magically work without a little prodding, beyond the hardware that *always* just magically works, of course. Asking for more is madness. -T.C. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 15:28, Orion Poplawski (or...@cora.nwra.com) wrote: On 05/15/2013 03:26 PM, Bill Nottingham wrote: Orion Poplawski (or...@cora.nwra.com) said: On 05/15/2013 05:36 AM, Lennart Poettering wrote: On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: 479ms iprupdate.service 385ms iprinit.service These appear to be untis that are only necessary for IBM RAID. It's hugely disappointing if this is pulled into all boots. I wonder if we can find a different logic for this, for example pulling it in from a udev rules or so. Or by changing anaconda to install this only onb IBM RAID systemd... Anyone knows what this is about? Also, I don't see either of these listed in the default preset file. This really shouldnt be started by default. 78ms iprdump.service IBM RAID again? Jeez... These are from the iprutils package: Summary : Utilities for the IBM Power Linux RAID adapters Description : Provides a suite of utilities to manage and configure SCSI devices supported by the ipr SCSI storage device driver. And are sysv init scripts: /etc/rc.d/init.d/iprdump /etc/rc.d/init.d/iprinit /etc/rc.d/init.d/iprupdate Which is listed a mandatory in the core group of comps. CCing iprutils-owner for comment. 99% sure those should only be built on PPC. Bill * Wed Sep 05 2012 Karsten Hopp kars...@redhat.com 2.3.11-1 - update to 2.3.11 - enable on all archs as it now supports some adapters on them, too. I filed a bug about this now: https://bugzilla.redhat.com/show_bug.cgi?id=963679 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/14/2013 06:30 PM, Dan Williams wrote: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: This is not intended to be snarky, but I admit it could sound like it is. When are long startup times for services considered to be bugs in their own right? [root@f19q ~]# systemd-analyze blame 1min 444ms sm-client.service 1min 310ms sendmail.service 18.602s firewalld.service 13.882s avahi-daemon.service 12.944s NetworkManager-wait-online.service Is anything waiting on NetworkManager-wait-online in your install? That target is really intended for servers where you want to block Apache or Sendmail or Database from starting until you're sure your networking is fully configured. If you don't have any services that require a network to be up, then you can mask NetworkManager-wait-online and things will be more parallel. 12.715s restorecond.service I have no idea why restorecond would take this much time, and it really should not be enabled for most machines, with file name transitions. Not sure why it is enabled. Could someone check on a fresh install if it is enabled? -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.13 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlGU2roACgkQrlYvE4MpobMWkgCePCWW4miPesJMTl3OyQIAue4r zu8AnR2iilICj6NqhjDfWZl4l7gMaXe6 =2nKR -END PGP SIGNATURE- -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/16/2013 07:10 AM, Daniel J Walsh wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/14/2013 06:30 PM, Dan Williams wrote: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: 12.715s restorecond.service I have no idea why restorecond would take this much time, and it really should not be enabled for most machines, with file name transitions. Not sure why it is enabled. Could someone check on a fresh install if it is enabled? policycoreutils-restorecond is marked mandatory in the gnome-desktop comp groups. I don't have it installed on my machines (KDE). -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA, Boulder/CoRA Office FAX: 303-415-9702 3380 Mitchell Lane or...@nwra.com Boulder, CO 80301 http://www.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 7:10 AM, Daniel J Walsh dwa...@redhat.com wrote: 12.715s restorecond.service I have no idea why restorecond would take this much time, and it really should not be enabled for most machines, with file name transitions. Not sure why it is enabled. Could someone check on a fresh install if it is enabled? That result was from a fresh install in a qemu-kvm VM. Fresh installed (wipefs -a the F19 partitions, then remove the partitions with fdisk, reinstall) on bare metal, and on the fifth reboot I get: 17.495s restorecond.service Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
These are the remaining items that I don't understand why they run at all, on every boot. There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service and no oops, so this seems unnecessary: 1.092s abrt-uefioops.service and I'm not using LVM so these seem unnecessary: 2.783s lvm2-monitor.service 489ms systemd-udev-settle.service 15ms lvm2-lvmetad.service How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 12:20 -0600, Chris Murphy wrote: These are the remaining items that I don't understand why they run at all, on every boot. There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service and no oops, so this seems unnecessary: 1.092s abrt-uefioops.service So abrt-uefioops.service always runs (it doesn't have any systemd conditionals), and it does this: # Wait for abrtd to start. Give it at least 1 second to initialize. i=10 while ! pidof abrtd /dev/null; do if test $((i--)) = 0; then exit 1 fi sleep 1 done sleep 1 cd /sys/fs/pstore 2/dev/null || exit 0 abrt-merge-uefioops -o * | abrt-dump-oops -D if test $? = 0; then abrt-merge-uefioops -d * fi Now, the ordering there looks wonky to me. It waits for abrtd to start and *then* checks if /sys/fs/pstore even exists. So I see various possible improvements here. 1. Why is it waiting for abrtd to 'initialize' at all? That seems like a messy hack. If abrtd.service is returning early, surely that should be fixed, rather than putting ugly sleep loops in things that depend on it. 2. Move the 'wait for abrtd' stuff to under the 'cd /sys/fs/pstore 2/dev/null || exit 0' line. Then at least if /sys/fs/pstore doesn't exist at all, we can exit without waiting for a second. 3. Make the whole service conditional on there being anything in /sys/fs/pstore at all. In practice, just doing this would resolve the problems and make 1 and 2 unnecessary (though still possibly desirable). To do that, just add this to the abrt-uefioops.service file: ConditionDirectoryNotEmpty=/sys/fs/pstore How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? Yes. 'rpm -qf /lib/systemd/system/foo.service'. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 12:20, Chris Murphy (li...@colorremedies.com) wrote: There have been no crashes, so ext4 doesn't need fsck on every boot: 4.051s systemd-fsck-root.service 515ms systemd-fsck@dev-disk-by\x2duuid-09c66d01\x2d8126\x2d39c2\x2db7b8\x2d25f14cbd35af.service Well, but only fsck itself knows that and can determine this from the superblock. Hence we have to start it first and it will then exit quickly if the fs wasn't dirty. Note that these times might be misleading: if fsck takes this long to check the superblock and exit this might be a result of something else which runs in parallel monopolizing CPU or IO (for example readahead), and might not actually be fsck's own fault. and no oops, so this seems unnecessary: 1.092s abrt-uefioops.service https://bugzilla.redhat.com/show_bug.cgi?id=963182 and I'm not using LVM so these seem unnecessary: 2.783s lvm2-monitor.service 489ms systemd-udev-settle.service 15ms lvm2-lvmetad.service How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? $ repoquery --qf=%{sourcerpm} --whatprovides '*/lib/systemd/system/lvm2-monitor.service' lvm2-2.02.98-8.fc19.src.rpm Please file a bug against the lvm2 package. And make sure to add it to: https://bugzilla.redhat.com/show_bug.cgi?id=963210 Hmm, on your machine, what does systemctl show -p WantedBy -p RequiredBy systemd-udev-settle.service show? This will tell us which package is actually responsible for pulling in systemd-udev-settle.service. Thanks! Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
From: Adam Williamson awill...@redhat.com How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? Yes. 'rpm -qf /lib/systemd/system/foo.service'. I'd actually suggest doing: rpm -qif /lib/systemd/system/foo.service ... and noting the source rpm. I've had some cases where the source rpm (bugzilla component) results in numerous binary rpms and sometimes the connections aren't so obvious at first. For example, note the Name vs. the Source RPM with: $ rpm -qif /usr/lib/python2.7/site-packages/imgcreate/fs.py Name: python-imgcreate Epoch : 1 Version : 18.15 Release : 1.fc18 Architecture: x86_64 Install Date: Wed 24 Apr 2013 11:47:37 AM EDT Group : System Environment/Base Size: 285483 License : GPLv2 Signature : RSA/SHA256, Fri 05 Apr 2013 01:34:00 AM EDT, Key ID ff01125cde7f38bd Source RPM : livecd-tools-18.15-1.fc18.src.rpm Build Date : Wed 03 Apr 2013 01:55:56 PM EDT Build Host : buildvm-13.phx2.fedoraproject.org Relocations : (not relocatable) Packager: Fedora Project Vendor : Fedora Project URL : http://git.fedorahosted.org/git/livecd Summary : Python modules for building system images Description : Python modules that can be used for building images for things like live image or appliances. -- John Florian -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 09:28, Orion Poplawski (or...@cora.nwra.com) wrote: On 05/16/2013 07:10 AM, Daniel J Walsh wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/14/2013 06:30 PM, Dan Williams wrote: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: 12.715s restorecond.service I have no idea why restorecond would take this much time, and it really should not be enabled for most machines, with file name transitions. Not sure why it is enabled. Could someone check on a fresh install if it is enabled? policycoreutils-restorecond is marked mandatory in the gnome-desktop comp groups. I don't have it installed on my machines (KDE). I have now filed a bug about this: https://bugzilla.redhat.com/show_bug.cgi?id=963919 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 14:44 -0400, john.flor...@dart.biz wrote: From: Adam Williamson awill...@redhat.com How do I determine what component to file a bug against? I guess I have to find the package that caused these .service files to be installed? Yes. 'rpm -qf /lib/systemd/system/foo.service'. I'd actually suggest doing: rpm -qif /lib/systemd/system/foo.service ... and noting the source rpm. I've had some cases where the source rpm (bugzilla component) results in numerous binary rpms and sometimes the connections aren't so obvious at first. For example, note the Name vs. the Source RPM with: Yeah, I usually do that in two steps, but doing it in one seems more logical. :) -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 12:39 PM, Lennart Poettering mzerq...@0pointer.de wrote: Hmm, on your machine, what does systemctl show -p WantedBy -p RequiredBy systemd-udev-settle.service show? This will tell us which package is actually responsible for pulling in systemd-udev-settle.service. RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service And I missed one other I don't understand, which is a top 3 offender: 9.855s accounts-daemon.service # We pull this in by graphical.target instead of waiting for the bus # activation, to speed things up a little: gdm uses this anyway so it is nice # if it Fine, but 10 seconds seems like a lot of time, but is it? Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 12:51 -0600, Chris Murphy wrote: And I missed one other I don't understand, which is a top 3 offender: 9.855s accounts-daemon.service # We pull this in by graphical.target instead of waiting for the bus # activation, to speed things up a little: gdm uses this anyway so it is nice # if it Fine, but 10 seconds seems like a lot of time, but is it? Honestly, it does seem like *everything* seems to start up slow on your F19 system, compared to F18. I wonder if there's some kind of bug causing general slowness on your system? On my desktop accounts-daemon.service takes 21ms, lvm2-monitor takes 70ms, lvm2-lvmetad 5ms, NetworkManager.service 454ms...the only thing that takes more than 1s is abrt-uefioops.service, and I explained why that was in my last mail. This is a fast system with an SSD, but even so, it seems like _everything_ is sluggish on your system. I think that's why someone asked if you were on a debug kernel. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 12:51, Chris Murphy (li...@colorremedies.com) wrote: On May 16, 2013, at 12:39 PM, Lennart Poettering mzerq...@0pointer.de wrote: Hmm, on your machine, what does systemctl show -p WantedBy -p RequiredBy systemd-udev-settle.service show? This will tell us which package is actually responsible for pulling in systemd-udev-settle.service. RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. Good to know though that LVM isn't the primary reason why we are slow. Now it's dmraid. And thank you, LVM guys, for getting rid of the udev-settle dep. Much appreciated! And I missed one other I don't understand, which is a top 3 offender: 9.855s accounts-daemon.service # We pull this in by graphical.target instead of waiting for the bus # activation, to speed things up a little: gdm uses this anyway so it is nice # if it Fine, but 10 seconds seems like a lot of time, but is it? Hmm, yeah, 10s is a lot, especially since this service is loaded kinda late and shouldn't really do anything. Matthias, any clue what this might be about? Does accounts-daemon do anything on start-up that might take long? Network stuff? Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 11:35 -0700, Adam Williamson wrote: 3. Make the whole service conditional on there being anything in /sys/fs/pstore at all. In practice, just doing this would resolve the problems and make 1 and 2 unnecessary (though still possibly desirable). To do that, just add this to the abrt-uefioops.service file: ConditionDirectoryNotEmpty=/sys/fs/pstore I'll submit a patch for abrt with this change after I test it, working on it ATM. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 12:15, Adam Williamson (awill...@redhat.com) wrote: On Thu, 2013-05-16 at 11:35 -0700, Adam Williamson wrote: 3. Make the whole service conditional on there being anything in /sys/fs/pstore at all. In practice, just doing this would resolve the problems and make 1 and 2 unnecessary (though still possibly desirable). To do that, just add this to the abrt-uefioops.service file: ConditionDirectoryNotEmpty=/sys/fs/pstore I'll submit a patch for abrt with this change after I test it, working on it ATM. Please reference this bug for the patch! Thanks! https://bugzilla.redhat.com/show_bug.cgi?id=963182 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service And I don't have a clue why the hell I've got an avahi-daemon.service running on every startup an why it take so extreme long to start. $ systemctl show -p WantedBy -p RequiredBy avahi-daemon.service RequiredBy= WantedBy=multi-user.target -- Regards Heiko Adams -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 21:29 +0200, Heiko Adams wrote: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service And I don't have a clue why the hell I've got an avahi-daemon.service running on every startup an why it take so extreme long to start. It's enabled by default because zeroconf is considered a useful thing to have. If you don't want it, you can happily disable it, I think. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Once upon a time, Heiko Adams heiko.ad...@gmail.com said: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service Some people seem to be seeing some extremely long startup times for some services. I can't see what would make some of these take that long. I'm wondering: - is there some common bug that is causing a major slowdown for some of these services? - is there possibly a bug in how systemd is measuring and/or reporting these times? Some of the services you list could be running into some type of network timeout, but tuned? restorecond? Those shouldn't be hitting the network AFAIK. Rather than focusing on individual services, it would seem to me like a good idea to see if there is some underlying issue at work here. -- Chris Adams li...@cmadams.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Am 16.05.2013 21:41, schrieb Chris Adams: Once upon a time, Heiko Adams heiko.ad...@gmail.com said: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service Some people seem to be seeing some extremely long startup times for some services. I can't see what would make some of these take that long. I'm wondering: - is there some common bug that is causing a major slowdown for some of these services? - is there possibly a bug in how systemd is measuring and/or reporting these times? Some of the services you list could be running into some type of network timeout, but tuned? restorecond? Those shouldn't be hitting the network AFAIK. Rather than focusing on individual services, it would seem to me like a good idea to see if there is some underlying issue at work here. For me the most anoying thing is the long delay of lightdm.service which causes a black screen for several seconds which could make me think something went wrong. But you're right it looks like a more general problem. -- Regards Heiko Adams -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 12:57 PM, Adam Williamson awill...@redhat.com wrote: On Thu, 2013-05-16 at 12:51 -0600, Chris Murphy wrote: And I missed one other I don't understand, which is a top 3 offender: 9.855s accounts-daemon.service # We pull this in by graphical.target instead of waiting for the bus # activation, to speed things up a little: gdm uses this anyway so it is nice # if it Fine, but 10 seconds seems like a lot of time, but is it? Honestly, it does seem like *everything* seems to start up slow on your F19 system, compared to F18. I wonder if there's some kind of bug causing general slowness on your system? No idea, as with the exception of the very high (5s to 1m) service times, I get really non-deterministic numbers for other services. For example on 10 reboots, colord.service is all over the map, with a high of 1.2 seconds and a low of 22ms! That's a huge spread. And it's not the only one. sshd.service is 23ms in the current baremetal boot, but has been as high as 12s. Massive differences. This is a fast system with an SSD, but even so, it seems like _everything_ is sluggish on your system. I think that's why someone asked if you were on a debug kernel. I'm not using a debug kernel. SSD makes a huge difference. Even Fedora19 on Vbox on OSX on SSD is faster than what I'm getting on baremetal with HDD. By a lot. [0.051627] [Firmware Bug]: ACPI: BIOS _OSI(Linux) query ignored [0.717606] [Firmware Bug]: efi: Inconsistent initial sizes Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 1:41 PM, Chris Adams li...@cmadams.net wrote: Once upon a time, Heiko Adams heiko.ad...@gmail.com said: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service Some people seem to be seeing some extremely long startup times for some services. I can't see what would make some of these take that long. I wonder if wireless connectivity is related? When I reboot with and without a wired connection attached, I get the same result. But maybe wireless is causing delays just by being configured, even if a wired connection is available? Rather than focusing on individual services, it would seem to me like a good idea to see if there is some underlying issue at work here. Is why I'm not trying to focus on individual services beyond understanding what they do or what they need, to see if there's a pattern. My top suspect is network related, because I get the weirdest changes by just changing hostname and rebooting, which is baffling to me. Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 11:35 -0700, Adam Williamson wrote: 2. Move the 'wait for abrtd' stuff to under the 'cd /sys/fs/pstore 2/dev/null || exit 0' line. Then at least if /sys/fs/pstore doesn't exist at all, we can exit without waiting for a second. 3. Make the whole service conditional on there being anything in /sys/fs/pstore at all. In practice, just doing this would resolve the problems and make 1 and 2 unnecessary (though still possibly desirable). To do that, just add this to the abrt-uefioops.service file: ConditionDirectoryNotEmpty=/sys/fs/pstore I've sent patches for both of these to the ABRT ML. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 1:51 PM, Chris Murphy li...@colorremedies.com wrote: I wonder if wireless connectivity is related? When I reboot with and without a wired connection attached, I get the same result. But maybe wireless is causing delays just by being configured, even if a wired connection is available? OK well this isn't right. On every reboot, wireless is activated and connected to. But the wired connection is disabled. I just removed the b43 wireless firmware files to prevent it from being involved, and on every wired reboot, Gnome has the wired connection set to off. I turn it on, I connect, I reboot, and it's off again. That's not how Live media works (it's on on every boot), and isn't how F18 works either. So I don't know what's going on… journalctl | grep p5p1 May 16 14:13:50 F19.localdomain systemd-udevd[187]: renamed network interface eth0 to p5p1 May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): carrier is OFF May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): new Ethernet device (driver: 'sky2' ifindex: 2) May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): exported as /org/freedesktop/NetworkManager/Devices/0 May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): device state change: unmanaged - unavailable (reason 'managed') [10 20 2] May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): bringing up device. May 16 14:14:06 F19.localdomain kernel: sky2 :0c:00.0 p5p1: enabling interface May 16 14:14:06 F19.localdomain kernel: IPv6: ADDRCONF(NETDEV_UP): p5p1: link is not ready May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): preparing device. May 16 14:14:06 F19.localdomain NetworkManager[403]: info (p5p1): deactivating device (reason 'managed') [2] May 16 14:14:08 F19.localdomain kernel: sky2 :0c:00.0 p5p1: Link is up at 100 Mbps, full duplex, flow control both May 16 14:14:08 F19.localdomain kernel: IPv6: ADDRCONF(NETDEV_CHANGE): p5p1: link becomes ready May 16 14:14:08 F19.localdomain NetworkManager[403]: info (p5p1): carrier now ON (device state 20) May 16 14:14:08 F19.localdomain NetworkManager[403]: info (p5p1): device state change: unavailable - disconnected (reason 'carrier-changed') [20 30 40] May 16 14:14:10 F19.localdomain avahi-daemon[308]: Registering new address record for fe80::21e:c2ff:fe1d:507e on p5p1.*. This is where I manually slide the Wired setting from Off to On in Gnome: May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) starting connection 'ens5' May 16 14:14:36 F19.localdomain NetworkManager[403]: info (p5p1): device state change: disconnected - prepare (reason 'none') [30 40 0] May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 1 of 5 (Device Prepare) scheduled... May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 1 of 5 (Device Prepare) started... May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 2 of 5 (Device Configure) scheduled... May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 1 of 5 (Device Prepare) complete. May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 2 of 5 (Device Configure) starting... May 16 14:14:36 F19.localdomain NetworkManager[403]: info (p5p1): device state change: prepare - config (reason 'none') [40 50 0] May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 2 of 5 (Device Configure) successful. May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 2 of 5 (Device Configure) complete. May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 3 of 5 (IP Configure Start) scheduled. May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 3 of 5 (IP Configure Start) started... May 16 14:14:36 F19.localdomain NetworkManager[403]: info (p5p1): device state change: config - ip-config (reason 'none') [50 70 0] May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Beginning DHCPv4 transaction (timeout in 45 seconds) May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Beginning IP6 addrconf. May 16 14:14:36 F19.localdomain avahi-daemon[308]: Withdrawing address record for fe80::21e:c2ff:fe1d:507e on p5p1. May 16 14:14:36 F19.localdomain NetworkManager[403]: info Activation (p5p1) Stage 3 of 5 (IP Configure Start) complete. May 16 14:14:36 F19.localdomain NetworkManager[403]: info (p5p1): DHCPv4 state changed nbi - preinit May 16 14:14:36 F19.localdomain dhclient[1526]: Listening on LPF/p5p1/00:1e:c2:1d:50:7e May 16 14:14:36 F19.localdomain dhclient[1526]: Sending on LPF/p5p1/00:1e:c2:1d:50:7e May 16 14:14:36 F19.localdomain dhclient[1526]: DHCPREQUEST on p5p1 to 255.255.255.255 port 67 (xid=0x5cc85c0c) May 16 14:14:36 F19.localdomain NetworkManager[403]: info (p5p1): DHCPv4
Re: when startup delays become bugs
Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Bill -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 16:17 -0400, Bill Nottingham wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. dmraid looks like it's only in the anaconda-tools group (and a critpath group) in comps, not in standard. But there's a possibly problematic dep chain here too: initial-setup - anaconda - python-blivet - dmraid so for non-GNOME systems, we're getting dmraid via that dep chain. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/16/2013 08:17 PM, Bill Nottingham wrote: We*could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? JBG -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 2013-05-16 at 20:41 +, Jóhann B. Guðmundsson wrote: On 05/16/2013 08:17 PM, Bill Nottingham wrote: We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? dmraid is hardly enterprise, though. It's used for all firmware RAID implementations aside from Intel's, and it's not unusual for any random system to be using firmware RAID. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On May 16, 2013, at 1:51 PM, Chris Murphy li...@colorremedies.com wrote: But maybe wireless is causing delays just by being configured, even if a wired connection is available? Not really. The one improvement is in NetworkManager-wait-online.service which with a wired connection instead of wireless is 1/10th of the time. So at the moment I'm still no farther along on finding the overall cause of slowness. 16.656s plymouth-quit-wait.service 13.750s accounts-daemon.service 12.924s avahi-daemon.service 12.852s chronyd.service 12.724s firewalld.service Chris Murphy -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 14:41, Chris Adams (li...@cmadams.net) wrote: Once upon a time, Heiko Adams heiko.ad...@gmail.com said: My top 6 of extreme long starting services are: $ systemd-analyze blame 27.652s NetworkManager.service 27.072s chronyd.service 27.015s avahi-daemon.service 26.899s tuned.service 26.647s restorecond.service 23.512s lightdm.service Some people seem to be seeing some extremely long startup times for some services. I can't see what would make some of these take that long. I'm wondering: - is there some common bug that is causing a major slowdown for some of these services? - is there possibly a bug in how systemd is measuring and/or reporting these times? Some of the services you list could be running into some type of network timeout, but tuned? restorecond? Those shouldn't be hitting the network AFAIK. Rather than focusing on individual services, it would seem to me like a good idea to see if there is some underlying issue at work here. So, the blame chart should not be misunderstood. It simply tells you how much time passed between the time systemd forked off the process until it completed initialization and told systemd about it. Now, within that time there might be many things happening and the service might simply wait for some resource to become available rather than be slow in its own. That resource could be the CPU or IO or some other service or device. For example, readahead might monopolize IO for some time during boot, so that other processes get starved. These dependencies and resource constraints are not visible in systemd-analyze blame. If you want to track this down, try systemd-bootchart. Simply boot with init=/usr/lib/systemd/systemd-bootchart, see systemd-bootchart(1) for details. It will plot your CPU and IO consumption and can show you resource usage by process, which is usually a more useful tool than a simple systemd-analyze blame. For the super slow run above I'd be quite interested to have a look at the bootchart actually. (Heiko? Can you upload that?) Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 13:44, Adam Williamson (awill...@redhat.com) wrote: On Thu, 2013-05-16 at 20:41 +, Jóhann B. Guðmundsson wrote: On 05/16/2013 08:17 PM, Bill Nottingham wrote: We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? dmraid is hardly enterprise, though. It's used for all firmware RAID implementations aside from Intel's, and it's not unusual for any random system to be using firmware RAID. As I understood this non-intel fakeraid is actually pretty much the exception... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Thu, 16.05.13 16:17, Bill Nottingham (nott...@redhat.com) wrote: Lennart Poettering (mzerq...@0pointer.de) said: RequiredBy= WantedBy=dmraid-activation.service I'm not using dmraid, or md raid, or any kind of raid at the moment. I also have this entry, previously explained in this thread as probably not being needed unless dmraid is being used, so is the likely offender for udev-settle. 2.823s dmraid-activation.service Given that this is really only needed for exotic stuff I do wonder why we need to install this by default even. We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Does anaconda have infrastructure for this? I'd certainly prefer if this could be done that way, even if it doesn't help livecd installs... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Fri, 2013-05-17 at 01:21 +0200, Lennart Poettering wrote: On Thu, 16.05.13 13:44, Adam Williamson (awill...@redhat.com) wrote: On Thu, 2013-05-16 at 20:41 +, Jóhann B. Guðmundsson wrote: On 05/16/2013 08:17 PM, Bill Nottingham wrote: We *could* drop all the assorted local storage tools from @standard and just leave them to be installed by anaconda if they're being used to create such storage. They'd have to remain on the live image, though, and so this would not help installs done from the live images. Is not the live images exactly the place where we dont want enterprise storage daemons? dmraid is hardly enterprise, though. It's used for all firmware RAID implementations aside from Intel's, and it's not unusual for any random system to be using firmware RAID. As I understood this non-intel fakeraid is actually pretty much the exception... Intel firmware RAID is becoming more popular, yeah, just as a result of the fact that Intel-based motherboards are becoming more popular; but there are still plenty of others out there. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Am 17.05.2013 01:09, schrieb Lennart Poettering: For the super slow run above I'd be quite interested to have a look at the bootchart actually. (Heiko? Can you upload that?) Lennart https://www.dropbox.com/s/95dzgklajrgaz4n/bootchart-20130517-0756.svg BTW: Is it correct that /dev/mapper/VolGroup-lv_home can't be found with systemd-bootchart? -- Regards Heiko Adams -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
2013/5/15 Chris Adams li...@cmadams.net: Once upon a time, Chris Murphy li...@colorremedies.com said: The only things related to networking I change is setting a hostname from localhost to f18s, f19s or f19q; and occasionally adding the b43 firmware if I decide to go wireless. Do you add the changed hostname to /etc/hosts? If not, sendmail (and some other things) will stop for some time trying to resolve the local hostname to an IP address. Why doesn't nss_myhostname resolve local hostname even if it is not in /etc/hosts? Alexey -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 12:56 AM, Jeffrey Bastian wrote: The slowest component on F19 is this new wait service: 10.102s NetworkManager-wait-online.service This service doesn't seem to do anything other than wait until the network is fully up. This really skews unfavorably the boot time reported by systemd-analyze. But if I mask that service and reboot, everthing still appears to work fine and systemd-analyze reports a boot time of only 6.2 seconds. A bit OT... This is not the first time in this thread that someone mentions masking a service. Do you guys have a reason why you mask the service instead of simply disabling it? Is there anything pulling the service into your boot even when it's disabled? Thanks, Michal -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Il 15/05/2013 05:43, Adam Williamson ha scritto: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: But firewalld goes from 7 seconds to 18 seconds? Why? avahi-daemon, restorecond, all are an order of magnitude longer on F19 than F18. It's a 3+ minute userspace hit on the startup time where the kernel takes 1.9 seconds. Off hand this doesn't seem reasonable, especially sendmail. If the time can't be brought down by a lot, can it ship disabled by default? FWIW, I found your results here interesting, so I did a little test of my own. I did default DVD installs of F17, F18 and F19 Beta TC4, ran through firstboot, rebooted, then rebooted again and ran systemd-analyze (to let prelink kick in). Results are that F18's slightly slower than F17 and F19 is somewhat slower again. My numbers are way way faster than your numbers overall, though; VMs do seem to perform very quickly on this box for whatever reason. Can you include here the QEMU command line or libvirt XML definition? Unless you have something like cache=none or cache='none' in it (respectively for QEMU and libvirt), chances are that you're using the host memory as a very fast and large disk cache for your VM. :) (VMs tend to be fast in the initramfs anyway, because they do not have much hardware to initialize, especially graphics cards). Paolo F17 --- Startup finished in 493ms (kernel) + 794ms (initramfs) + 2751ms (userspace) = 4039ms 446ms udev-settle.service 345ms NetworkManager.service 268ms systemd-logind.service 266ms ip6tables.service 262ms avahi-daemon.service 261ms iptables.service 173ms mcelog.service 170ms nfs-lock.service 145ms udev-trigger.service 136ms udev.service 135ms abrt-ccpp.service 122ms spice-vdagentd.service 117ms sendmail.service 114ms sm-client.service 110ms media.mount 106ms sys-kernel-debug.mount 105ms fedora-loadmodules.service 105ms dev-hugepages.mount 103ms dev-mqueue.mount 100ms sys-kernel-security.mount 98ms rsyslog.service 91ms remount-rootfs.service 90ms dbus.service 86ms sys-kernel-config.mount 84ms systemd-vconsole-setup.service 84ms acpid.service 77ms boot.mount 62ms systemd-tmpfiles-setup.service 56ms abrt-vmcore.service 53ms fedora-storage-init.service 53ms systemd-user-sessions.service 49ms sshd.service 47ms auditd.service 44ms systemd-remount-api-vfs.service 41ms colord.service 41ms systemd-sysctl.service 38ms bluetooth.service 32ms fedora-storage-init-late.service 28ms fedora-readonly.service 28ms lvm2-monitor.service 27ms systemd-readahead-collect.service 25ms colord-sane.service 18ms udisks2.service 18ms mdmonitor-takeover.service 13ms upower.service 13ms accounts-daemon.service 11ms rtkit-daemon.service 10ms rpcbind.service 9ms fedora-wait-storage.service F18 --- Startup finished in 521ms (kernel) + 616ms (initramfs) + 3348ms (userspace) = 4485ms 742ms iscsid.service 607ms firewalld.service 460ms systemd-udev-settle.service 372ms chronyd.service 369ms restorecond.service 321ms gdm.service 292ms abrt-ccpp.service 279ms ksmtuned.service 231ms accounts-daemon.service 208ms spice-vdagentd.service 208ms auditd.service 182ms systemd-logind.service 174ms avahi-daemon.service 167ms rtkit-daemon.service 163ms sm-client.service 116ms fedora-readonly.service 110ms fedora-loadmodules.service 109ms systemd-udev-trigger.service 104ms NetworkManager.service 101ms mcelog.service 95ms systemd-udevd.service 87ms sendmail.service 84ms sys-kernel-debug.mount 82ms dev-hugepages.mount 82ms dev-mqueue.mount 79ms iscsi.service 74ms systemd-remount-fs.service 64ms sys-kernel-config.mount 56ms systemd-vconsole-setup.service 52ms colord.service 42ms fedora-storage-init.service 41ms systemd-user-sessions.service 35ms udisks2.service 34ms ksm.service 34ms polkit.service 29ms systemd-tmpfiles-setup.service 28ms rpcbind.service 26ms bluetooth.service 24ms fedora-storage-init-late.service 23ms sshd.service 16ms abrt-vmcore.service 15ms upower.service 15ms lvm2-monitor.service 15ms systemd-sysctl.service 13ms systemd-modules-load.service 13ms mdmonitor-takeover.service 10ms boot.mount 4ms tmp.mount F19 --- Startup finished in 411ms (kernel) + 745ms (initrd) + 4.704s (userspace) = 5.861s 2.745s plymouth-quit-wait.service 2.389s NetworkManager-wait-online.service 1.078s accounts-daemon.service 1.026s firewalld.service 1.007s restorecond.service 987ms avahi-daemon.service 479ms iprupdate.service 385ms iprinit.service 356ms systemd-udev-settle.service
Re: when startup delays become bugs
On Tue, 14.05.13 17:58, Chris Murphy (li...@colorremedies.com) wrote: Static hostname: f19q.localdomain Takes 1.5 minutes before I can ssh into the VM, but the sendmail.service is now about 2 seconds instead of a minute. Is this a non-debug kernel? [root@f19q ~]# systemd-analyze blame 51.688s firewalld.service 50.696s restorecond.service 16.989s avahi-daemon.service 16.023s systemd-logind.service 11.908s NetworkManager-wait-online.service With the recommendations from https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 NM-w-o.s would become a service that is only pulled in when actually needed. It would be good to get this included in F19, as it is completely bogus to delay the boot for everybody just because a few people happen to have static NFS mounts listed in fstab. BTW, systemd-analyze critical-chain (possibly with --fuzz=500ms) is a new tool in F19, which allows a bit more Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Tue, 14.05.13 17:30, Dan Williams (d...@redhat.com) wrote: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: This is not intended to be snarky, but I admit it could sound like it is. When are long startup times for services considered to be bugs in their own right? [root@f19q ~]# systemd-analyze blame 1min 444ms sm-client.service 1min 310ms sendmail.service 18.602s firewalld.service 13.882s avahi-daemon.service 12.944s NetworkManager-wait-online.service Is anything waiting on NetworkManager-wait-online in your install? That target is really intended for servers where you want to block Apache or Sendmail or Database from starting until you're sure your networking is fully configured. If you don't have any services that require a network to be up, then you can mask NetworkManager-wait-online and things will be more parallel. Dan, could you please implement these recommendations for F19: https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 This really shouldn't be pulled in unconditionally... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Tue, 14.05.13 15:51, Chris Murphy (li...@colorremedies.com) wrote: 2.911s abrt-uefioops.service I don't understand really why abrt needs to run at all by default. It should be spawned automatically when an actualy crash happens, so that by default nobody has to have this serice running. Abrt guys, any chance you can make use of socket/bus/path activation so that abrt is auto-spawned when needed and not before? You don't even need to patch abrt itself for that (dropping in unit files is sufficient) if it needs only bus or path activation. Only socket activation needs patching. I filed this now, to keep track of this: https://bugzilla.redhat.com/show_bug.cgi?id=963184 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: Let me take the opportunity to dissect some of this: Startup finished in 411ms (kernel) + 745ms (initrd) + 4.704s (userspace) = 5.861s 2.745s plymouth-quit-wait.service This one is misleading, as this just makes sure the bootsplash is terminated when start-up is complete. 2.389s NetworkManager-wait-online.service https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 479ms iprupdate.service 385ms iprinit.service These appear to be untis that are only necessary for IBM RAID. It's hugely disappointing if this is pulled into all boots. I wonder if we can find a different logic for this, for example pulling it in from a udev rules or so. Or by changing anaconda to install this only onb IBM RAID systemd... Anyone knows what this is about? Also, I don't see either of these listed in the default preset file. This really shouldnt be started by default. 356ms systemd-udev-settle.service This is exlusively LVM's fault. There's really no need to ever have that in the boot unless you run LVM. On many machines LVM is the primary reason why things are slow. I can only recommend everybody to not install on LVM if you can. It's a real shame that LVM hasn't gotten their things together still, after all those years. A service that needs systemd-udev-settle is a broken service! LVM, you are broken! I am hoping for the day LVM gets kicked out of the default install. Oh btrfs, why are you still not around? 297ms ksmtuned.service I thought the kernel did this without user-interaction these days? 236ms spice-vdagentd.service I don't see why this is on by default. According to this bug report this should have been pulled in on-demand only: https://bugzilla.redhat.com/show_bug.cgi?id=876237 233ms abrt-ccpp.service https://bugzilla.redhat.com/show_bug.cgi?id=963184 160ms fedora-loadmodules.service Somebody should file bugs against all packages that still drop something into /etc/sysconfig/modules/. I see that kvm and bluez do this still. I filed these bugs now: https://bugzilla.redhat.com/show_bug.cgi?id=963193 https://bugzilla.redhat.com/show_bug.cgi?id=963198 138ms sm-client.service Oh my, sendmail. What an emabrassement that we still ship this enabled by default... If we could at least turn it into something that is activated on-demand. But I gues touching these sources to implement that would be too awful... 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. 97ms sshd.service Dito. THis is something to start by default only on hosts where a ton of people log in all the time. 80ms isdn.service In very new systemd 'isdn.service' is not enabled any more in the preset file, so this is already gone now. 79ms systemd-modules-load.service Ah, cute, on my machine the only reason this is pulled in is /etc/modules-load.d/spice-vdagentd.conf -- which also pulls in our old friend uinput (see above). https://bugzilla.redhat.com/show_bug.cgi?id=963201 78ms iprdump.service IBM RAID again? Jeez... 64ms sendmail.service Urks... 56ms systemd-localed.service Hmm, this one is weird... It should only be loaded when needed, so I am surprised tjat there's something that needs this during boot-up. 54ms ksm.service I thought the kernel would do this internally these days without userspace interaction. 54ms abrt-vmcore.service https://bugzilla.redhat.com/show_bug.cgi?id=963184 45ms rpcbind.service https://bugzilla.redhat.com/show_bug.cgi?id=963189 33ms dmraid-activation.service I wonder if we can change this to pull this in only if DMRAID is actually used... SOunds wrong to spawn this unconditionally... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 12:07, Alexey Torkhov (atork...@gmail.com) wrote: 2013/5/15 Chris Adams li...@cmadams.net: Once upon a time, Chris Murphy li...@colorremedies.com said: The only things related to networking I change is setting a hostname from localhost to f18s, f19s or f19q; and occasionally adding the b43 firmware if I decide to go wireless. Do you add the changed hostname to /etc/hosts? If not, sendmail (and some other things) will stop for some time trying to resolve the local hostname to an IP address. Why doesn't nss_myhostname resolve local hostname even if it is not in /etc/hosts? My suspicion is that sendmail uses res_query() and friends since it needs MX and stuff... THis means NSS is bypassed and hence nss_myhostname and /etc/hosts won't take effect anyway... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 01:36:57PM +0200, Lennart Poettering wrote: On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: 356ms systemd-udev-settle.service This is exlusively LVM's fault. There's really no need to ever have that in the boot unless you run LVM. On many machines LVM is the primary reason why things are slow. I can only recommend everybody to not install on LVM if you can. It's a real shame that LVM hasn't gotten their things together still, after all those years. Is there a constructive summary anywhere of what LVM needs to do / change? This problem affects libguestfs too. I am hoping for the day LVM gets kicked out of the default install. Oh btrfs, why are you still not around? I Want To Believe in btrfs, but unfortunately it's still excessively buggy. It's actually got worse in Rawhide recently -- it reliably fails in mkfs.btrfs :-( I stopped bothering filing bugs about this because they just get ignored because Rawhide isn't new enough (they want us to retest everything on some non-upstream btrfs-next kernel). My colleague ran btrfs on his laptop and had daily crashes. Last week he went back to ext4. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://libguestfs.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 2013-05-15 at 13:36 +0200, Lennart Poettering wrote: On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: Let me take the opportunity to dissect some of this: Startup finished in 411ms (kernel) + 745ms (initrd) + 4.704s (userspace) = 5.861s 2.745s plymouth-quit-wait.service This one is misleading, as this just makes sure the bootsplash is terminated when start-up is complete. 2.389s NetworkManager-wait-online.service https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 479ms iprupdate.service 385ms iprinit.service These appear to be untis that are only necessary for IBM RAID. It's hugely disappointing if this is pulled into all boots. I wonder if we can find a different logic for this, for example pulling it in from a udev rules or so. Or by changing anaconda to install this only onb IBM RAID systemd... Anyone knows what this is about? Also, I don't see either of these listed in the default preset file. This really shouldnt be started by default. 356ms systemd-udev-settle.service This is exlusively LVM's fault. There's really no need to ever have that in the boot unless you run LVM. On many machines LVM is the primary reason why things are slow. I can only recommend everybody to not install on LVM if you can. It's a real shame that LVM hasn't gotten their things together still, after all those years. A service that needs systemd-udev-settle is a broken service! LVM, you are broken! I am hoping for the day LVM gets kicked out of the default install. Oh btrfs, why are you still not around? 297ms ksmtuned.service I thought the kernel did this without user-interaction these days? 236ms spice-vdagentd.service I don't see why this is on by default. According to this bug report this should have been pulled in on-demand only: https://bugzilla.redhat.com/show_bug.cgi?id=876237 233ms abrt-ccpp.service https://bugzilla.redhat.com/show_bug.cgi?id=963184 160ms fedora-loadmodules.service Somebody should file bugs against all packages that still drop something into /etc/sysconfig/modules/. I see that kvm and bluez do this still. I filed these bugs now: https://bugzilla.redhat.com/show_bug.cgi?id=963193 https://bugzilla.redhat.com/show_bug.cgi?id=963198 138ms sm-client.service Oh my, sendmail. What an emabrassement that we still ship this enabled by default... If we could at least turn it into something that is activated on-demand. But I gues touching these sources to implement that would be too awful... 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. 97ms sshd.service Dito. THis is something to start by default only on hosts where a ton of people log in all the time. 80ms isdn.service In very new systemd 'isdn.service' is not enabled any more in the preset file, so this is already gone now. 79ms systemd-modules-load.service Ah, cute, on my machine the only reason this is pulled in is /etc/modules-load.d/spice-vdagentd.conf -- which also pulls in our old friend uinput (see above). https://bugzilla.redhat.com/show_bug.cgi?id=963201 78ms iprdump.service IBM RAID again? Jeez... 64ms sendmail.service Urks... 56ms systemd-localed.service Hmm, this one is weird... It should only be loaded when needed, so I am surprised tjat there's something that needs this during boot-up. 54ms ksm.service I thought the kernel would do this internally these days without userspace interaction. 54ms abrt-vmcore.service https://bugzilla.redhat.com/show_bug.cgi?id=963184 45ms rpcbind.service https://bugzilla.redhat.com/show_bug.cgi?id=963189 33ms dmraid-activation.service I wonder if we can change this to pull this in only if DMRAID is actually used... SOunds wrong to spawn this unconditionally... Lennart -- Lennart Poettering - Red Hat, Inc. Lennart, good. I CC'ed for interested bugs. -- Best Regards, Igor Gnatenko -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 01:36:57PM +0200, Lennart Poettering wrote: 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. There are few parts of iSCSI stack in Fedora. This iscsi.service brings up all targets previously configured. Nevertheless it should only be pulled in when there are actaully some targets discovered: https://bugzilla.redhat.com/show_bug.cgi?id=951951#c3 Other part of stack, iscsid.service, is already socket activable (at least upstream since Mike pulled my patches, in Fedora since 6.2.0.873-4). -- Tomasz Torcz ,,If you try to upissue this patchset I shall be seeking xmpp: zdzich...@chrome.pl an IP-routable hand grenade.'' -- Andrew Morton (LKML) -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 13:36, Lennart Poettering (mzerq...@0pointer.de) wrote: I have now created a tracker bug WhyWeBotSoSlow to track all these little issues that cause things to be in our boot-up sequence that really shouldn't be there... https://bugzilla.redhat.com/show_bug.cgi?id=963210 Would be fantastic if we could find some volunteers to keep an eye on this and put some pressure behind this! Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 12:55, Richard W.M. Jones (rjo...@redhat.com) wrote: On Wed, May 15, 2013 at 01:36:57PM +0200, Lennart Poettering wrote: On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: 356ms systemd-udev-settle.service This is exlusively LVM's fault. There's really no need to ever have that in the boot unless you run LVM. On many machines LVM is the primary reason why things are slow. I can only recommend everybody to not install on LVM if you can. It's a real shame that LVM hasn't gotten their things together still, after all those years. Is there a constructive summary anywhere of what LVM needs to do / change? This problem affects libguestfs too. It should subscribe to block devices coming and going and then assemble things as stuff appears. Right now it expects to be called at a point in time where everything that might show up has shown up. Of course, that's not how modern hardware works (everything is hotpluggable now, hw gives no guarantees when it shows up and we don't have any idea what might still come hence), and requires pulling in of udev-settle. It requires us to delay the boot, in order to keep the promise of everything has shown up now to LVM, which happens to be a promise that we cannot actually hold, since we don't know whether something is missing... LVM should subscribe to udev events like any other daemon, and as soon as all devices it needs have shown up immeidately assemble things. This would guarantee minimal boot times since we would have to wait exactly for the devices that are actually needed, and that's it. We wouldn't have to wait for devices we never know might appear and that nobody actually needs. If it wasn't for LVM we'd have removed the udev-settle functionality from udev already, since it's just broken, and slows down things too. LVM of course has been broken like this for 5 years, and they know that. And they did nothing about it. And that's so disappointing... (Of course, btrfs got assembly right from day 1: it will assemble disks exclusively via udev events and as soon as a device is complete it is made available to the OS. The btrfs folks understood how modern hardware and storage technology works, the LVM folks not so much I fear...) I am hoping for the day LVM gets kicked out of the default install. Oh btrfs, why are you still not around? I Want To Believe in btrfs, but unfortunately it's still excessively buggy. It's actually got worse in Rawhide recently -- it reliably fails in mkfs.btrfs :-( I stopped bothering filing bugs about this because they just get ignored because Rawhide isn't new enough (they want us to retest everything on some non-upstream btrfs-next kernel). My colleague ran btrfs on his laptop and had daily crashes. Last week he went back to ext4. Well, it works fine for myself and for quite a few other folks I know. I am pretty sure it's time to grow some, make this the default in Fedora, and then fix everything popping up quickyl with the necessary pressure. As it appears not having any pressure to stabilize btrfs certainly doesn't work at all for the project... Lennart PS: I am very good at bitching about LVM. You can book me for your party at very low rates! -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 02:25 PM, Lennart Poettering wrote: On Wed, 15.05.13 12:55, Richard W.M. Jones (rjo...@redhat.com) wrote: Is there a constructive summary anywhere of what LVM needs to do / change? This problem affects libguestfs too. It should subscribe to block devices coming and going and then assemble things as stuff appears. They have added the lvmetad daemon which does this. And it seems to be used by default on my test F19 installation. I.e. I am not seeing udev-settle being pulled in anymore. Michal -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 2013-05-15 at 12:52 +0200, Lennart Poettering wrote: On Tue, 14.05.13 17:30, Dan Williams (d...@redhat.com) wrote: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: This is not intended to be snarky, but I admit it could sound like it is. When are long startup times for services considered to be bugs in their own right? [root@f19q ~]# systemd-analyze blame 1min 444ms sm-client.service 1min 310ms sendmail.service 18.602s firewalld.service 13.882s avahi-daemon.service 12.944s NetworkManager-wait-online.service Is anything waiting on NetworkManager-wait-online in your install? That target is really intended for servers where you want to block Apache or Sendmail or Database from starting until you're sure your networking is fully configured. If you don't have any services that require a network to be up, then you can mask NetworkManager-wait-online and things will be more parallel. Dan, could you please implement these recommendations for F19: https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 This really shouldn't be pulled in unconditionally... The changes you suggest there seem fine; one question about After=syslog.target being removed though: intentional? I pretty much just take your guidance on the .service files. Dan -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 07:45, Dan Williams (d...@redhat.com) wrote: Is anything waiting on NetworkManager-wait-online in your install? That target is really intended for servers where you want to block Apache or Sendmail or Database from starting until you're sure your networking is fully configured. If you don't have any services that require a network to be up, then you can mask NetworkManager-wait-online and things will be more parallel. Dan, could you please implement these recommendations for F19: https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 This really shouldn't be pulled in unconditionally... The changes you suggest there seem fine; one question about After=syslog.target being removed though: intentional? I pretty much just take your guidance on the .service files. After=syslog.target is unnecessary these days and should be removed from all unit files, to keep things simple. We nowadays require syslog implementations to be socket activatable, and that socket is around before normal services start, and that's guaranteed, hence nobody has to depend on syslog explicitly anymore. All services will have logging available anyway, out-of-the-box, as default, with no manual configuration necessary, and without any referring to syslog.target. Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 02:48 PM, Lennart Poettering wrote: On Wed, 15.05.13 07:45, Dan Williams (d...@redhat.com) wrote: Is anything waiting on NetworkManager-wait-online in your install? That target is really intended for servers where you want to block Apache or Sendmail or Database from starting until you're sure your networking is fully configured. If you don't have any services that require a network to be up, then you can mask NetworkManager-wait-online and things will be more parallel. Dan, could you please implement these recommendations for F19: https://bugzilla.redhat.com/show_bug.cgi?id=787314#c37 This really shouldn't be pulled in unconditionally... The changes you suggest there seem fine; one question about After=syslog.target being removed though: intentional? I pretty much just take your guidance on the .service files. After=syslog.target is unnecessary these days and should be removed from all unit files, to keep things simple. We nowadays require syslog implementations to be socket activatable, and that socket is around before normal services start, and that's guaranteed, hence nobody has to depend on syslog explicitly anymore. All services will have logging available anyway, out-of-the-box, as default, with no manual configuration necessary, and without any referring to syslog.target. is it documented? in many of your mails in this thread you wrote nowadays, anymore, default, modern system, modern hardware, etc... does it documented anywhere? what are the assumptions and how nowadays fedora should have to work? -- Levente Si vis pacem para bellum! -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 02:58 PM, Farkas Levente wrote: On 05/15/2013 02:48 PM, Lennart Poettering wrote: We nowadays require syslog implementations to be socket activatable, and that socket is around before normal services start, and that's guaranteed, hence nobody has to depend on syslog explicitly anymore. All services will have logging available anyway, out-of-the-box, as default, with no manual configuration necessary, and without any referring to syslog.target. is it documented? in many of your mails in this thread you wrote nowadays, anymore, default, modern system, modern hardware, etc... does it documented anywhere? what are the assumptions and how nowadays fedora should have to work? The requirements on syslog implementations to be compatible with systemd are described in: http://www.freedesktop.org/wiki/Software/systemd/syslog Michal -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 01:01:18PM +0200, Lennart Poettering wrote: On Tue, 14.05.13 15:51, Chris Murphy (li...@colorremedies.com) wrote: 2.911s abrt-uefioops.service I don't understand really why abrt needs to run at all by default. It should be spawned automatically when an actualy crash happens, so that by default nobody has to have this serice running. This specific one is scraping previous kernel crash reports out of pstore, so it should really only be doing something if there's anything there. -- Matthew Garrett | mj...@srcf.ucam.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Hi, On 05/15/2013 01:36 PM, Lennart Poettering wrote: On Tue, 14.05.13 20:43, Adam Williamson (awill...@redhat.com) wrote: Let me take the opportunity to dissect some of this: snip 236ms spice-vdagentd.service I don't see why this is on by default. According to this bug report this should have been pulled in on-demand only: Adam is running a spice enabled vm, so it is getting started because he has the required virtual hardware. https://bugzilla.redhat.com/show_bug.cgi?id=876237 233ms abrt-ccpp.service https://bugzilla.redhat.com/show_bug.cgi?id=963184 160ms fedora-loadmodules.service Somebody should file bugs against all packages that still drop something into /etc/sysconfig/modules/. I see that kvm and bluez do this still. I filed these bugs now: https://bugzilla.redhat.com/show_bug.cgi?id=963193 https://bugzilla.redhat.com/show_bug.cgi?id=963198 79ms systemd-modules-load.service Ah, cute, on my machine the only reason this is pulled in is /etc/modules-load.d/spice-vdagentd.conf -- which also pulls in our old friend uinput (see above). https://bugzilla.redhat.com/show_bug.cgi?id=963201 I'm fine with dropping the config file from spice-vdagent as long as something then creates the /dev/uinput device node, so that module auto-load will actually work, without the device-node pre-existing trying to use /dev/uinput will just cause a -ENOENT failure. snip 54ms ksm.service I thought the kernel would do this internally these days without userspace interaction. At a minimum we should consider not starting these *inside* vms Regards, Hans -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Once upon a time, Lennart Poettering mzerq...@0pointer.de said: 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. This is attaching to configured iSCSI devices (which at a minimum requires parsing configuration files to see if there are any devices configured), not running a listening daemon. 97ms sshd.service Dito. THis is something to start by default only on hosts where a ton of people log in all the time. SSH host key generation needs to be done in advance (don't want a connecting socket to wait for that). Maybe that could be done with a separate firstboot-like service that gets disabled once run? -- Chris Adams li...@cmadams.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Tue, 2013-05-14 at 22:21 -0600, Chris Murphy wrote: I've tried both, but when I rename it I'm using 'hostnamectl set-hostname XXX' as I'm not seeing anything obvious in Gnome that does this. This is on the Details panel of System Settings. You're expected to type a pretty hostnames with spaces and caps. You'll end up with something like [michael@victory-road ~]$ hostnamectl Static hostname: victory-road Pretty hostname: Victory Road It doesn't magically become fully qualified. signature.asc Description: This is a digitally signed message part -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 08:53, Chris Adams (li...@cmadams.net) wrote: Once upon a time, Lennart Poettering mzerq...@0pointer.de said: 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. This is attaching to configured iSCSI devices (which at a minimum requires parsing configuration files to see if there are any devices configured), not running a listening daemon. It should be possible to come up with some form of ConditionPathExists= or ConditionDirectoryNotEmpty= that causes this to be skipped if no targets are configured. https://bugzilla.redhat.com/show_bug.cgi?id=951951 97ms sshd.service Dito. THis is something to start by default only on hosts where a ton of people log in all the time. SSH host key generation needs to be done in advance (don't want a connecting socket to wait for that). Maybe that could be done with a separate firstboot-like service that gets disabled once run? We should really to be as stateless as possible here and not require write access to /etc, which a solution like this would require. Instead I'd propose to splitt the key generation into its own service but then pull this in by the first connection and conditionalize it also with ConditionPathExists= or so: ConditionPathExists=!/etc/ssh/ssh_host_rsa_key I filed this now: https://bugzilla.redhat.com/show_bug.cgi?id=963268 Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 15:38, Hans de Goede (hdego...@redhat.com) wrote: Ah, cute, on my machine the only reason this is pulled in is /etc/modules-load.d/spice-vdagentd.conf -- which also pulls in our old friend uinput (see above). https://bugzilla.redhat.com/show_bug.cgi?id=963201 I'm fine with dropping the config file from spice-vdagent as long as something then creates the /dev/uinput device node, so that module auto-load will actually work, without the device-node pre-existing trying to use /dev/uinput will just cause a -ENOENT failure. Yeah, kmod/udev/systemd get this right these days: the kernel modules just export in modalias what node they want to have created, kmod then extracts that and udev/systemd create the device node for it. Then, when userspace opens the deivce the kernel will make sure the module is actually auto-loaded. Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 2013-05-15 at 11:58 +0200, Paolo Bonzini wrote: Il 15/05/2013 05:43, Adam Williamson ha scritto: On Tue, 2013-05-14 at 15:51 -0600, Chris Murphy wrote: But firewalld goes from 7 seconds to 18 seconds? Why? avahi-daemon, restorecond, all are an order of magnitude longer on F19 than F18. It's a 3+ minute userspace hit on the startup time where the kernel takes 1.9 seconds. Off hand this doesn't seem reasonable, especially sendmail. If the time can't be brought down by a lot, can it ship disabled by default? FWIW, I found your results here interesting, so I did a little test of my own. I did default DVD installs of F17, F18 and F19 Beta TC4, ran through firstboot, rebooted, then rebooted again and ran systemd-analyze (to let prelink kick in). Results are that F18's slightly slower than F17 and F19 is somewhat slower again. My numbers are way way faster than your numbers overall, though; VMs do seem to perform very quickly on this box for whatever reason. Can you include here the QEMU command line or libvirt XML definition? Unless you have something like cache=none or cache='none' in it (respectively for QEMU and libvirt), chances are that you're using the host memory as a very fast and large disk cache for your VM. :) That's likely the case, yeah, I'm using a standard VM definition, no tweaks, and I have 16GB of RAM. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 2013-05-15 at 13:36 +0200, Lennart Poettering wrote: 236ms spice-vdagentd.service I don't see why this is on by default. According to this bug report this should have been pulled in on-demand only: https://bugzilla.redhat.com/show_bug.cgi?id=876237 It probably was. I was testing in a SPICE/qxl KVM. Please don't anyone take it out; I really appreciate having it in all my test VMs. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 08:53:44AM -0500, Chris Adams wrote: SSH host key generation needs to be done in advance (don't want a connecting socket to wait for that). Maybe that could be done with a separate firstboot-like service that gets disabled once run? Bugzilla # ? :) -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 10:57:52AM -0400, Matthew Miller wrote: On Wed, May 15, 2013 at 08:53:44AM -0500, Chris Adams wrote: SSH host key generation needs to be done in advance (don't want a connecting socket to wait for that). Maybe that could be done with a separate firstboot-like service that gets disabled once run? Bugzilla # ? :) Oh -- Lennart is ahead of me. (mumbles about the good advice of reading all the messages before sending replies) -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 04:21:20PM +0200, Lennart Poettering wrote: https://bugzilla.redhat.com/show_bug.cgi?id=963268 In that bug you suggest that administrators could change back to standalone mode. Given that what you say about frequency of access is true (even on systems where ssh login is a primary activity, time between accesess is generally on a human scale rather hundreds per second), is there a disadvantage to socket activation beyond a very slight delay in the first access? -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Once upon a time, Matthew Miller mat...@fedoraproject.org said: In that bug you suggest that administrators could change back to standalone mode. Given that what you say about frequency of access is true (even on systems where ssh login is a primary activity, time between accesess is generally on a human scale rather hundreds per second), is there a disadvantage to socket activation beyond a very slight delay in the first access? Well, SSH connection rate is on human scale until your system is the target of a scan, which can result in at least dozens per second. -- Chris Adams li...@cmadams.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, May 15, 2013 at 10:05:40AM -0500, Chris Adams wrote: Once upon a time, Matthew Miller mat...@fedoraproject.org said: In that bug you suggest that administrators could change back to standalone mode. Given that what you say about frequency of access is true (even on systems where ssh login is a primary activity, time between accesess is generally on a human scale rather hundreds per second), is there a disadvantage to socket activation beyond a very slight delay in the first access? Well, SSH connection rate is on human scale until your system is the target of a scan, which can result in at least dozens per second. Oh, I see. I missed that the idea is to activate sshd each time rather than as a singleton service. -- Matthew Miller ☁☁☁ Fedora Cloud Architect ☁☁☁ mat...@fedoraproject.org -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On 05/15/2013 02:21 PM, Lennart Poettering wrote: On Wed, 15.05.13 08:53, Chris Adams (li...@cmadams.net) wrote: Once upon a time, Lennart Poettering mzerq...@0pointer.de said: 112ms iscsi.service This really sounds like something that should be socket actviated on demand rather than run by default. This is attaching to configured iSCSI devices (which at a minimum requires parsing configuration files to see if there are any devices configured), not running a listening daemon. It should be possible to come up with some form of ConditionPathExists= or ConditionDirectoryNotEmpty= that causes this to be skipped if no targets are configured. https://bugzilla.redhat.com/show_bug.cgi?id=951951 97ms sshd.service Dito. THis is something to start by default only on hosts where a ton of people log in all the time. SSH host key generation needs to be done in advance (don't want a connecting socket to wait for that). Maybe that could be done with a separate firstboot-like service that gets disabled once run? We should really to be as stateless as possible here and not require write access to /etc, which a solution like this would require. Instead I'd propose to splitt the key generation into its own service but then pull this in by the first connection and conditionalize it also with ConditionPathExists= or so: ConditionPathExists=!/etc/ssh/ssh_host_rsa_key I filed this now: https://bugzilla.redhat.com/show_bug.cgi?id=963268 I created such service ( ssh-keygen.service ) when I migrated the ssh to unit files. For some reason the maintainers in the distribution chose not to include it.. JBG -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
On Wed, 15.05.13 10:05, Chris Adams (li...@cmadams.net) wrote: Once upon a time, Matthew Miller mat...@fedoraproject.org said: In that bug you suggest that administrators could change back to standalone mode. Given that what you say about frequency of access is true (even on systems where ssh login is a primary activity, time between accesess is generally on a human scale rather hundreds per second), is there a disadvantage to socket activation beyond a very slight delay in the first access? Well, SSH connection rate is on human scale until your system is the target of a scan, which can result in at least dozens per second. Note that systemd puts a (configurable) limit on the number of concurrent connections, much like sshd does it, so there's very little difference... Also ssh forks of per-connection processes too, so the difference is probably even smaller... Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: when startup delays become bugs
Once upon a time, Lennart Poettering mzerq...@0pointer.de said: Note that systemd puts a (configurable) limit on the number of concurrent connections, much like sshd does it, so there's very little difference... Also ssh forks of per-connection processes too, so the difference is probably even smaller... The difference is that when sshd is running as a traditional daemon, it can do all of its start-up processing once (parsing config files, loading host keys, etc.), instead of for each connection. -- Chris Adams li...@cmadams.net -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel