[00:01] I also agree with Daekdroom, why not just use apt-get for everything in place of Aptitude? [00:02] foxmulder881, I actually believe it should be the other way around. [00:02] foxmulder881: err, we meant use aptitude and get rid of apt-get [00:02] Daekdroom; sorry. Misunderstood. I guess we disagree then! lol [00:03] heh, I hate aptitude's menu crap [00:03] and always use apt-get [00:03] Daekdroom; what's so special about Aptitude? [00:03] foxmulder881: why do you think we should use apt-get? because it has less features? [00:03] foxmulder881, it's "smarter" [00:03] patdk-wk: you don't have to use the curses interface... [00:03] patdk-wk, I hate its interface, but the command line is awesome [00:03] yofel; it does everything required and is simple to use and understand. I personally can't stand Aptitude. [00:04] * yofel actually likes the interface... [00:04] errk.. [00:04] foxmulder881: err... how are the command line options of aptitude more complicated than the apt-get ones? [00:04] patdk-wk: you ever used aptitude via cli, no ncrusers? [00:05] its more powerful then apt-get [00:05] never knew you could :) [00:05] and haven't seen anything that told me you could [00:05] I actually found out recently aptitude had a ncurses interface. [00:05] Tripped over it by accident a few days ago :P [00:05] everything is aptitude menu stuff, or apt-get [00:05] patdk-wk: aptitude install BLA [00:05] or my fab [00:06] alias aptitudeupgrade='sudo aptitude update ; sudo aptitude safe-upgrade ; sudo aptitude full-upgrade' [00:06] patdk-wk: aptitude --help [00:06] So what can Aptitude do that apt-get can't? [00:06] Genuine question. [00:06] aptitude moo [00:06] foxmulder881: resolve some dep messes , for starter? [00:06] intelligent dependency solution suggestions [00:07] citations needed! [00:07] yeah that too [00:07] not that I ever use it [00:07] jcastro: ? [00:07] "resolve some dep messes , for starter? " [00:08] jcastro: I felt in love with aptitude when we did the kde 3.5->4.x migration [00:08] The gain is more like 2MB by the way (tasksel+aptitude), and ONLY if the install is done via ubiquity, d-i will still use it. [00:08] nothing does package dep resolving better then aptitude [00:08] By dep messes I'm assuming you mean general dep error for which apt-get can also fix? Or different. [00:08] apt-get can fix dep fails? [00:08] never did for me [00:08] specially when you have to downgrade a package [00:09] So if you use alternate install, you'll always get aptitude, if I've understood things correctly [00:09] -f install can I think, but it never did what I wanted [00:09] lol [00:09] I stop using -f IONs ago [00:09] Yes, -f is what I'm referring to. [00:09] I don't know what you mean by "dep fail" [00:09] never actually fixes anythig [00:09] Works for me! [00:10] and I'm not sure if apt-get will show you more than one resolution [00:10] I thought -f only fixed the apt-get database of installed crap [00:10] but not fix anything [00:10] jcastro: do you run devel cicle before beta? :p [00:10] Hmm, it has fixed countless things for me, but anyways... [00:10] BUGabundo: yes, for the past 5 years! [00:10] then you should know what I me [00:10] take the X migration going right now [00:10] aptitude won't install anything critical of it [00:11] apt's been holding it back for me [00:11] apt-get full-upgrade would, no questions asked [00:11] you mean dist-upgrade? [00:11] What the hell is full-upgrade? [00:12] yes sorry [00:12] the habit of aptitude [00:12] lol [00:12] i've been running apt-get upgrade every day and it's holding back the packages like it's supposed to [00:12] Apt-get and Aptitude breed different types of lingo! [00:12] Well, same questions as aptitude full-upgrade, really, but aptitude tries to keep it sane, and tells you things, apt-get just says here's what I'll do, dowant? [00:12] they do [00:12] dist-upgrade is for upgrading from one distro release to another [00:12] one of the most confusing one [00:13] is aptitude install also does upgrades [00:13] arand; that's crap. [00:14] just tried to compare safe-upgrade with apt-get upgrade and they already disagree, aptitude will install the new kernel images, apt-get won't [00:14] \o/ [00:14] * BUGabundo serves a slice of bread with peanut butter to yofel [00:14] yofel; so find out why. [00:15] thanks [00:15] BUGabundo, aptitude install also removes "useless" packages, instead of only suggesting it like apt-get does [00:15] foxmulder881: no reason, apt-get holds the linux meta packages back, while aptitude simply installs the new files they now depend on and upgrades them [00:15] There's always a reason behind it if apt-get fails for some reason. I like to manually get behind the source of problems like that. I don't like applications to do everything for me and not let me know why it went pear-shaped in the first place. [00:15] I would love to see apt be more verbose some times [00:15] then again, we have dselect [00:15] apt installed the new kernel for me yesterday [00:15] jcastro: Hmm, well, it is for operations other than upgrading packages, e.g. removals and new-installs [00:16] Daekdroom: that's debian/ubuntu policy [00:16] has been for one or two cycles [00:16] BUGabundo, but apt-get doesn't do that at all. [00:16] It's aptitude specific [00:16] Which I guess once released, is rather irrelevant [00:16] did last time I tried on debian [00:17] Daekdroom: apt-get autoremove does. [00:17] arand, yeah, but aptitude does it with only an aptitude install. No specific command for that. [00:17] You always have to be more specific with apt-* no holding hands like with aptitude.. [00:17] if that was true, based on what synaptic shows as obsolete [00:18] I would have 500 MBs extra free space :) [00:18] arand; that's what I'm trying to say. That's the way I prefer things. [00:18] So we need a package manager that's verbose-esque like aptitude and specific like apt-get? [00:18] I guess so! [00:19] Yes, so do I, but I've come to understand that ubuntu will kill my babies, and such is life. [00:19] Daekdroom: dselect [00:20] or go all hardcore, and dpgk [00:20] *dpkg [00:20] don't use dselect, it will just make you want to kill yourself [00:20] I also like dpkg. But only for purging stuff and if no other deps required. But it is rock soild every time. [00:20] * patdk-wk uses dpkg* alot [00:20] jcastro: I was jk, ofc [00:21] heh [00:21] * yofel only uses dpkg if there are .debs to install or --force-* is needed [00:21] * Daekdroom doesn't like --force at all [00:22] it's there for a reason, and no, you shouldn't use it without knowing what you're doing [00:22] Hmm, building your own packages means a lot of dpkg, and dpkg -L/-S is just crazy useful [00:22] +1 [00:23] (forgot about that ^^) [00:23] BUGabundo: what are you doing tomorrow? Want to help with unity/global menu? [00:24] I'm off for now. Cheers folks. Off to reinstall lucid actually! [00:24] That tasksel is going though... not awfully sad about that, it's a beast of messing up a system.. [00:25] jcastro: last day of work, before one week vacations mostly offline [00:25] if weather permits [00:26] arand: tasksel is nice [00:26] jcastro: I really should re-test unity [00:26] haven't boot into it since it was out [00:26] BUGabundo: IT IS NOT: Bug #574287 [00:26] Launchpad bug 574287 in tasksel (Ubuntu) "tasksel uninstalled my system!" [Undecided,Confirmed] https://launchpad.net/bugs/574287 [00:26] holla at me when you want to try, I need help testing the global menu! [00:26] arand: LOOOL [00:26] jcastro: okay [00:27] I'll probably _cry_ about the here and StatusNet/twitter [00:27] I even created the SN group for it ) [00:27] :) [00:27] identi.ca/group/unity [00:28] I still need to test it out on debian, since presumably it's not an ubuntu change inducing that, seems like tasksel and ubuntu-desktop = unfriends... [00:28] ahaha [00:29] only if you clear -desktop task [00:29] duh [00:30] BUGabundo: Nope: 1 install openssh, 2 uninstall openssh = removal of half of gnome :/ [00:31] What does tasksel do? [00:31] what?!?! [00:31] no can do! [00:32] Daekdroom: Install a group of packages meant for a specific task. [00:32] Daekdroom: think like this [00:32] you want to install a LAMP [00:32] it does it all for you [00:32] no need to fetch, apache, mysql, php, etc by hand [00:33] its like a special meta-package [00:33] sooooooo [00:33] anyone can confirm mocp is DOA? [00:34] ..google says museum of contemporary photography, hmm.. [00:35] BUGabundo: mocp who? [00:52] well that stinks, no more 64 bit flash [00:53] Sarvatt elaborate please? [00:54] they closed down the beta since it was based on 10 instead of 10.1 [00:54] can't download it from their website anymore [01:12] pokity poke [01:26] Hi [01:30] Flash crashing in chrome? [01:36] lucitu: Has been for ages, I think.. [01:36] arand: only recently [01:36] Hmm, well I'm not a chromium user, but I rember it doing that a week or so ago... [01:37] arand: about 2wks ago when libgtk2.0 was upgraded..had to downgrade to lucid version to work [01:38] arand: looks like libgtk2.0 will stay that way.. [01:38] Sarvatt: Hmm, I've got the .so if you want it? [01:39] arand: nah I've got it, thank you though. just a shame they stopped it [01:40] lucitu: Hmm, yea, I'm sorry i don't think I can be much help there, but search LP, as is my general-handwavey-response to most things.. [01:40] I'm off to bed.. [01:49] who can I poke about alsa development? Rumour has it that the latest alsa has support for some speciality recording hardware that I have, but I haven't had any luck compiling it for 2.6.>28, and it appears that lucid's alsa doesn't have it, even though as far as I can tell it _should_ have it [01:49] m-audio fast track ultra, 8x8 channel usb [01:49] cwillu just compile the latest and see if it works [01:49] that's all [01:50] thanks for reading what I wrote :p [01:50] I mean on the later kernels [01:50] 2.6.>28 == anything newer than 2.6.28 [01:50] did you try 34 35 ? [01:50] yes [01:50] missing symbols [01:51] while compiling ? [01:51] but I can rebuild the kernel itself fine [01:51] yes [01:51] on all kernels or some specifics ? [01:52] not dead sure I've tried against 2.6.35 yet, hang on (that machine is in the other room, and I haven't installed ssh on it yet) [01:53] 2.6.34 is the latest I've tried [01:53] did you try to compile om maverick or lynx ? [01:53] well 35 is in the mainstream ppa rc 2 I believe [01:53] rc2 isn't, no; last I checked it only has a source deb [01:53] * Volkodav was about to install it too [01:53] http://kernel.ubuntu.com/~kernel-ppa/mainline/ [01:54] planning on moving to it on a couple servers as soon as rc3 is available (.35 has a bunch of btrfs fixes that I want, and rc2 has a known btrfs corruption issue) [01:54] all 3 debs are there for maverick only on 35 rc2 [01:54] oh, wait, I was looking at the lucid source, yep [01:54] Did you read about the regression on 35 ? [01:55] there's so many, did you mean a particular one? [01:55] overall performance speed etc [01:55] udev and btrfs are what I'm aware of [01:55] ya, that's a udev issue [01:55] that phoeronix junk [01:55] I would like to tinker btrfs on my SSD [01:55] afaik, that's already been sorted out [01:56] you can set up btrfs via dkms easily enough, just haven't bothered yet [01:56] _requires_ 2.6.35 currently though [01:56] is it in the installer option on maverick yet ? [01:56] no idea; [01:56] set up you mean convert ? [01:56] I've been using it for about a year though [01:56] * psusi is quite pleased with ext4 on his ssd... and on the new 1.5 tb wd drive [01:57] anyone wanna test e2defrag? ;) [01:57] no, I meant to install btrfs modules from git via dkms instead of what's in mainline (given that rc2 has some issues there) [01:57] cwillu did you try the snapshot yet on it ? [01:57] long ago mate :) [01:57] oh ok [01:57] I've been deploying embedded stuff on it for about 8 months [01:57] * psusi hopes to get his patch for e2fsck into maverick and lazy_itable_init on by default so it only takes 30 seconds instead of 15 minutes to format a 1.5 tb ext4 partition [01:58] checksumming is so very handy when you only have crappy sd cards as root drives [01:58] how is it different from ZFS dtrace ? [01:58] haven't used zfs [01:58] although my impression is that btrfs is on much more solid theoretical footing than zfs [01:59] we're probably going to have lzo compression in a day or two, raid5 is coming along nicely, etc [01:59] well then you have an option of playing with rc2 ot compile your own I guess [01:59] anyways, back to alsa :p [01:59] unless you wanna keep talking btrfs :p [01:59] how does raid5 plays with ssd s though ? [01:59] how do you mean? [01:59] I heard they had issues on early stages [02:00] they simly did not work [02:00] which raid5? It's not done yet [02:00] don't work is not an error description... [02:00] any raid for that matter with ssd [02:00] lots of things don't simply don't work, because they haven't been implemented [02:00] no, that's not true [02:00] drives don't know or care if they are in a raid [02:01] you're not going to be able to TRIM them currently though... [02:01] but internally every ssd is already a raid [02:01] layering btrfs on top of lvm has been troublesome, mainly because lvm didn't pass through barriers properly to the hardware [02:01] Volkodav, so are most harddrives [02:01] no, an ssd is an ssd [02:02] raid stands for redundant array of inexpensive disks... definitionally impossible for a single disk to be one [02:02] every ssd has internal raid controller with bunch of memory in it [02:02] Volkodav, it's more complicated than that in fact [02:02] no, it has a flash memory controller [02:02] the controller does much more than most raid controllers do [02:02] that emulates a sata disk [02:02] does trimming etc [02:03] no, trimming is something the os does [02:03] or is working towards doing... [02:03] wear levelling, compaction, etc [02:03] os kernel and FS [02:03] takes 3 just to trim [02:03] the performance differences between ssd drives largely comes down to the algorithms in use [02:03] os driver = part of kernel [02:03] err, fs driver rather [02:04] cwillu but the bottom line that they are all internally in raid [02:04] the companies with experience with incremental garbage collection were the first to get decent lasting performance [02:04] Volkodav, no, the bottom line is that they're apples and oranges [02:05] you might as well say a system with multiple mounts is a raid [02:05] apples ssd who is oranges ? [02:05] raid [02:05] or a disk with multiple heads... multiple heads do not a raid make... a floppy disk is not a raid... [02:06] so there is no difference in setting up ssd or regular harddrive say in raid [02:07] no, there's big differences [02:07] that's all I was saying [02:07] * psusi needs to do a thorough check to make sure everything is in order and release a new rev of e2defrag and get someone to sponsor uploading it to maverick universe [02:07] ... and we keep telling you you're wrong :p [02:07] people could not even boot them first [02:07] Volkodav, ssd drives? yes, yes they could [02:08] Before they had ata interfaces in front of them, they weren't at all like block devices at all [02:08] while raid is a way of taking a bunch of block devices, and ganging them into a single device [02:09] they just needed boot parameter dmraid=true [02:09] well back to alsa [02:09] :p [02:09] dmraid has nothing to do with ssd.. dmraid is bios fake hardware raid support [02:09] what are you trying to get to work ? [02:09] really only exists for compatibility with winders [02:10] piece of hardware called "m-audio fast track ultra" [02:10] I can get usb id's and such [02:10] currently only the midi interfaces comes up properly [02:10] hrm.. that reminds me... I need to poke someone about getting the dmraid regression in lucid SRU'd in time for the 10.04.1 respin... [02:10] so it is a newer sound card I guess ? [02:10] Somebody posted usb traces of all the mixer functionality under windows, I was planning on hacking that in, but I need to start with the basic support [02:11] it's a year or so old, but the others in its family also weren't supported [02:11] there's been patches floating around, but they're in the changelog for the latest alsa stable [02:11] and in the source [02:11] hmm [02:11] may as well patch [02:12] * Volkodav just patched exaile [02:12] mini mode is not working [02:12] was not* [02:12] Volkodav, I finding it hard to not say this in a rude fashion [02:12] Volkodav, but your reading comprehension leaves a little bit to be desired :p [02:13] oh all that was for windoze ? [02:13] * Volkodav getting focus together [02:13] case in point :p [02:13] let's start over [02:13] hmm [02:13] (a) alsa mainline doesn't compile against recent kernels without patches that I can't find [02:13] does luke chat in here? [02:14] (b) ubuntu's alsa doesn't seem to be recent enough to include the new drivers [02:14] do they exist those patches yet ? [02:14] (c) I need to be able to compile from some source, so that I can generate patches to send upstream [02:14] I would check gentoo forums [02:14] ooouf [02:15] and I will not under any circumstances be installing gentoo on my machines :p [02:15] you may find the patches on their forums [02:15] I did twice [02:15] I was hoping to find an alsa maintainer or someone with similar experience, to basically copy their approach to this [02:16] Volkodav, the problem isn't so much finding a way to make it build, and more finding the _right_ way to make it build [02:17] if I include patches to build it under 2.6.35, and those don't match what we in debian/ubuntu use, then it makes it harder for me to get support with testing and development, let alone upstreaming [02:17] oh we talking not just the dirty hack to make it work - you want elegant ? [02:17] The point of the endeavour is to not have magic packages on my system :) [02:17] or the alsa type approach [02:18] the sooner I can upstream it, the sooner I can work on _other_ things which need fixing :) [02:18] there must some channel for alsa I think [02:18] an ubuntu/debian alsa channel? [02:19] debian probably be a better shoice [02:19] or deb-dev [02:19] they have a few dev channels for debian [02:22] you know, I really wish I would remember sooner alternate nicks that particular people have :p [02:22] crimsun_, can I poke you for a couple minutes? [03:14] if anyone cares to test the defrag package, a new build is now in my ppa [03:18] psusi: wud love to..where is the ppa? am assuming the dev has to be unmounted? [03:19] bukayoo, on launchpad... ppa:psusi/ppa... yes, has to be unmounted, and of course, have a current backup, or otherwise don't care if the data on it is lost... works well for me, but still has a serious possibility of totally destroying all your data ;) [03:20] hello! [03:20] Am I the only one with rhytmbox plugins problems? [03:21] psusi: aside from destroyed data, how can I tell if success? does the % non-contiguous from fsck enough? [03:21] bukayoo, that's one indicator... it has an ncurses interface that shows the disk map too, and has a frag program to check fragmentation of specific files... also you can run the e2freegrag from e2fsprogs to check free space fragmentation [03:22] psusi: guessing it shud be zero after? [03:22] also run an e2fsck -f after defrag [03:22] yep [03:22] psusi: ok..will try [03:23] it's a little slow creating the relocation map now... changed it to use a double binary tree indexed extent list instead of a double array of block pointers so it can now handle very large disks without many, many gigs of ram, at the expense of some cpu [03:23] will need to do some optimizations to regain startup speed but I can now defrag a 1.5 tb disk with only 2 gigs of ram [03:24] ohh, you should also do a read only pass first with -r to make sure it works [03:25] and depending ont he size of the disk and ram size you might want to add -p and specify a larger buffer pool size than the default 32768 to speed things up [03:26] psusi: can I try on a vbox disk? [03:26] sure [03:27] psusi: great..i have some junk I can try. [03:27] for my testing I usually just make a test lvm volume and either restore a dump to it or dd a snapshot of my root volume [03:34] and in case the name led you to think otherwise, it does work on ext4 [03:34] now [03:34] psusi launchpad link please? [03:36] Sensiva, launchpad.net/e2defrag is the project... for the ppa just apt-add-repository ppa:psusi/ppa [03:36] psusi: could not determine the filesystem type. pls run the appropriate defragmenter [03:37] bukayoo, ohh... I forgot it even had a generic auto defragger... just run e2defrag instead... I'll have to take a look at the wrapper now [03:38] psusi: ok..now working [03:38] it used to support ext, xia, and minix too [03:41] oh weird... looks like the wrapper is just a script that calls file on the dev node and parses its output to decide what the fs is... that doesn't work on actual devices... anymore... [03:44] psusi: it's working now..just like pacman..will try later on a real hd bec the virtual disk is clean === [Nitsuga] is now known as Nitsuga [03:44] hehe [03:44] good old ascii graphics ;) [03:45] takes you back doesn't it? I remember using this thing when I first tried slackware back in '95 [03:46] psusi: yes it does..I see some 'B'ad blocks. what is that? [03:46] probably the journal [03:46] B is just generic for anything that can't be moved [03:47] psusi: ok [03:47] no, wait... I made the journal movable.... [03:47] hrm.... [03:47] how big of a volume is this? is it freshly formatted? ext4? it might be the resize inode... [03:47] but usually that's so small it doesn't show up [03:48] psusi: it's an ext4 [03:48] psusi: where is the image paste bin? [03:48] photobucket? [03:50] * psusi wishes Ted, Alex, and Linus hand't given up on this thing [03:50] err, Remmy not Alex [03:50] psusi: here is a snapshot http://imagebin.ca/view/0ABy22N6.html [03:50] ahh, yea, that's the resize inode [03:51] reserves a little space after each block group descriptor table for expansion [03:53] psusi: what are the little white dots? looks like it didn't put them all together? [03:53] free space, see the key? ;) [03:53] bundled closer to the front? [03:53] or legend [03:53] psusi: ohh..my bad [03:53] by default it tries to locate data in the block group of the owning inode [03:54] if you use the inode priority file option though to assign a high priority to an inode, it will force it to the start of the disk rather than in its native block group [03:55] this is handy for packing all the boot files ureadahead reads at the start of the disk, even if their inodes are scattered around [03:56] psusi: if I remember it correctly the windowz defragmenter kind of move them up front..was it windows or norton? [03:58] psusi: will try now on a real hd.. [03:59] psusi: but I think mine boots so fast some stuff aren't ready in time.. always get a fatal: module.dep load failure 3x before it loads properly [04:02] afaik, microsoft licensed diskkeeper, and that just kind of randomly moves fragmented files around until they are not fragmented anymore [04:03] this acts like it started with a completely empty disk, allocating blocks one after the other from left to right until every file on the disk has had blocks allocated... then moves everything from its current position to the new position [04:05] but like I said, it tries to allocate blocks in the same block group as the inode that wants them so they are located near the owning inode [04:06] psusi: just for my enlightenment.. I was thinking b4 can I achieve the same if I copy the whole disk out, reformat old disk and then copy back? [04:10] bukayoo, doing that with ext4 will generally result in low fragmentation, but does not do the same thing since the kernel uses different allocation algorithms [04:12] psusi: e2freefrag count error after - http://imagebin.ca/view/kOdpoy.html [04:13] run e2fsck -f first after defrag [04:14] right now defrag does not update the free blocks count in the superblock if it ends up allocating/freeing blocks for the extent tree so it can end up wrong... fsck will fix that easily.. that's on my todo list [04:15] psusi: here is the e2freefrag after http://imagebin.ca/view/xdeqNNm6.html [04:16] yep, that looks about what I've come to expect [04:17] psusi: here is the before - http://imagebin.ca/view/KgLIOZ.html [04:17] you can get larger than 2gb free space chunks if you up the flex_bg factor when you mkfs too with mke2fs -G x... the default is 32 which gives the 2gb maxium contiguous free space [04:18] psusi: ok..but remember this is a vbox virtual disk [04:19] created by vbox [04:21] psusi: anyway thnks for the nice tool..will try on real hd. will let you know [04:21] make sure you backup first ;) [04:24] have the xserver-xorg-video-* files been kept back for a while now? I've just installed so am curious if it's been like this for a while already? (I'm happily using "sudo aptitude safe-upgrade") [04:24] and it was originally written by Theodore Tso, Remmy Card, and Linus Torvalds... they just abandoned it ages ago and I'm bringing it back to life ;) [04:25] of course it will probably be rendered obsolete in a year or two by e4defrag, but hey... was interesting to work on and gives us something to use in the mean time [04:26] and hell, I might even port it to work on ntfs [04:26] and fat32 [04:27] psusi: definitely..what is e4defrag? the newer version from these trio? [04:27] defragger supposedly being written from scratch to do online defrag of ext4 [04:27] using ioctls built into the kernel rather than directly mucking with the disk [04:28] it has the advantage of being able to work on a mounted fs, and be safe by going through the journal... but since it has to allocate a new file, then atomically swap blocks, I doubt it will ever be as fast or be able to pack files as well as e2defrag [04:28] or optimize free space [04:29] if you loose power in the middle of an e2defrag, your screwed ;) [04:29] psusi: i wud think so.unless you stop all procs writing to it [04:31] psusi: i used to work with tandem and had to stop procs so to increase max extents of file near max [04:31] the really nifty thing about e2defrag is that you can build a priority list from the ureadahead pack file and pass it to e2defrag so it packs all those files sequentially at the start of the disk to speed up boot time... I don't see e4defrag ever being able to do that [04:33] ext4's delayed allocation, extents, and multiblock allocator really does a fantastic job of minimizing fragmentation [04:34] even for bit torrenting, which usually results in horrible fragmentation... ext4 keeps it contiguous as long as the torrent app uses fallocate() to tell it how big the file is going to be up front... [04:35] and even if it does't, ext4 still seems to do a pretty good job [04:35] psusi, there's no reason an online defragmenter couldn't do all the same things though [04:36] with the possible exception of dealing with a near full filesystem, [04:36]