| 00:25:50 | | AlsoHP_Archivist joins |
| 00:29:23 | | HP_Archivist quits [Ping timeout: 252 seconds] |
| 01:00:37 | | chrismeller (chrismeller) joins |
| 01:01:00 | | chrismeller quits [Remote host closed the connection] |
| 01:02:07 | | chrismeller (chrismeller) joins |
| 01:02:30 | | chrismeller quits [Remote host closed the connection] |
| 01:03:06 | | froschgrosch_ quits [Ping timeout: 240 seconds] |
| 01:03:37 | | chrismeller (chrismeller) joins |
| 01:04:00 | | chrismeller quits [Remote host closed the connection] |
| 01:04:12 | | AlsoHP_Archivist quits [Ping timeout: 265 seconds] |
| 01:05:07 | | chrismeller (chrismeller) joins |
| 01:05:30 | | chrismeller quits [Remote host closed the connection] |
| 01:05:53 | | chrismeller (chrismeller) joins |
| 01:06:51 | | AlsoHP_Archivist joins |
| 01:16:17 | | thuban quits [Read error: Connection reset by peer] |
| 01:16:38 | | thuban joins |
| 01:23:25 | | anarcat (anarcat) joins |
| 01:23:50 | <anarcat> | JAA: i don't know about you, but i'm actually getting tired of the loong raid/luks/lvm/ext4 stack |
| 01:24:02 | <anarcat> | JAA: i setup a ZFS server recently, and i absolutely love the lightweight snapshots |
| 01:24:13 | <anarcat> | i'm probably going slowly convert everything over |
| 01:24:30 | <anarcat> | only problem is requiring an external kernel module, because it's not mainline (yet?) in linux |
| 01:25:00 | <@JAA> | Heh, yeah, although it's also nice to have these things isolated from each other. |
| 01:25:18 | <@JAA> | But yes, ZFS is wonderful. |
| 01:25:30 | <anarcat> | true |
| 01:25:35 | <anarcat> | except when you need snapshots |
| 01:25:42 | <anarcat> | then it feels everything kind of falls over |
| 01:25:43 | <@JAA> | And yeah, not in mainline and probably never will be due to licence issues. |
| 01:25:50 | <anarcat> | performance goes to hell, and you never have enough space |
| 01:26:01 | <anarcat> | unless you uses thin provisionning, which i have never quite figured out |
| 01:26:11 | <anarcat> | oh, and SSD caching is a total mystery to me |
| 01:26:21 | <anarcat> | i set it up with LVM and it was *so* odd and confusing and scary |
| 01:26:34 | <anarcat> | with ZFS, that's another thing that's basically a box you check and it's done |
| 01:26:43 | <anarcat> | but yeah, probably never mainline unfortunately |
| 01:26:56 | <britmob|m> | I like Ceph. Gives me a lot of the benefits of ZFS and does a lot for multi-host setups. Works great for all my VM disks, and bulk archival storage is a good use case too. |
| 01:26:57 | | froschgrosch_ joins |
| 01:27:45 | <anarcat> | https://en.wikipedia.org/wiki/OpenZFS#Ubuntu_integration is actually interesting though |
| 01:27:54 | <@JAA> | I'd like to see btrfs in a better state, but at this point, I doubt they'll ever get RAID 5/6 stable. |
| 01:27:55 | <anarcat> | seems like ubuntu is shipping precompiled zfs modules from their repositories |
| 01:29:24 | <anarcat> | i just don't trust btrfs |
| 01:29:26 | <@JAA> | They are, but then you have Ubuntu. :-P |
| 01:29:35 | <anarcat> | i use it for one filesystem, without raid, and i regret it |
| 01:29:42 | <britmob|m> | Stupid Oracle, sticking their fingers in everything great :( |
| 01:29:46 | <anarcat> | i have no idea what's going on, it's hard to even tell what my partitions are |
| 01:30:02 | <anarcat> | JAA: well if they dodged the legal challenges, it opens the door for other distros to go that way too |
| 01:30:12 | <anarcat> | britmob|m: this actually predates oracle buying sun |
| 01:31:19 | <britmob|m> | oh yeah? interesting |
| 01:31:30 | <anarcat> | afaik |
| 01:31:48 | <britmob|m> | well I suppose oracle is continuing to make it an issue :P |
| 01:32:02 | <anarcat> | https://en.wikipedia.org/wiki/ZFS |
| 01:33:46 | <anarcat> | it was published by sun under the CDDL in 2005 |
| 01:34:03 | <anarcat> | sun was acquired in 2010 |
| 01:36:00 | <anarcat> | it's such a mess |
| 01:36:03 | <anarcat> | https://en.wikipedia.org/wiki/Common_Development_and_Distribution_License |
| 01:36:10 | <anarcat> | it's not even something Oracle would sue on |
| 01:36:20 | <@JAA> | Oracle could easily relicence their code, but all the other contributions would get complicated. |
| 01:36:22 | <anarcat> | CDDL is opensource, it's just incompatible with the GPL, but not according to everyone |
| 01:36:33 | <anarcat> | it's according to Eben Moglen and the Software Freedom Law Center |
| 01:37:06 | <anarcat> | and they didn't even explain why until Ubuntu actually sought legal advice and decided it was legal to ship the modules |
| 01:37:30 | <anarcat> | *then* moglen went out to say that it's actually legal if you follow "the letter of the law" but not "the spirit of the license" and that the latter would be enforcable in court |
| 01:37:34 | <anarcat> | it's kind of ridiculous |
| 01:38:28 | <anarcat> | honestly i think the biggest challenge in mainlining zfs at this point is not legal |
| 01:38:38 | <anarcat> | it's the hill they would have to climb to get the code accepted in mainline |
| 01:38:41 | <anarcat> | ted tso would throw a fit |
| 01:38:43 | <anarcat> | first off |
| 01:39:02 | <anarcat> | then the btrfs people would have a heart attack and collectively resign in protest |
| 01:39:24 | <anarcat> | *then* maybe it would get merged, but then you have the problem of syncing that source code across mac, the BSDs, and the linux kernel |
| 01:39:45 | <anarcat> | right now, openzfs is a common codebase for everyone, and it took years to get to that point... i'm not sure mainlining it would be an improvement |
| 01:43:07 | <@JAA> | Yeah, that's true. |
| 01:43:48 | <anarcat> | now i want to convince debian to do this |
| 01:43:55 | <anarcat> | good job, self, you just gave yourself another crusade |
| 01:44:19 | <@JAA> | :-) |
| 01:44:57 | <Wayward> | Can it be used in Windows, yet? |
| 01:45:04 | <anarcat> | so i heard |
| 01:45:42 | <@JAA> | Apparently: https://openzfsonwindows.org/ |
| 01:45:51 | <Wayward> | out of the box? |
| 01:45:55 | <anarcat> | https://en.wikipedia.org/wiki/OpenZFS#Microsoft_Windows |
| 01:46:11 | <anarcat> | 2.5admins.com say that it's a thing now |
| 01:46:24 | <anarcat> | but also, in the same breath, basically say "why on earth would you do something like this" basically |
| 01:46:29 | <anarcat> | that, and also it's not as stable |
| 01:46:36 | <anarcat> | which, you know, kind of matters for filesystems |
| 01:47:21 | <Wayward> | I have anxiety about storing 100 TB to a filesystem that's not time tested it won't destroy data. |
| 01:47:44 | <Wayward> | or someone plugs it into a computer and windows asks to format it |
| 01:48:26 | <@JAA> | Backups :-) |
| 01:48:59 | <anarcat> | which, in zfs, you tend to do with... zfs :p |
| 01:50:11 | <@JAA> | I wouldn't trust it that far, but yeah, sure. |
| 01:51:32 | <anarcat> | well that's what i mean, if you're worried about the filesystem stability, then you'll need another backup system |
| 01:51:43 | <anarcat> | this is actually one of the reasons i'm looking at zfs |
| 01:51:51 | <anarcat> | as part of a multi-layered backup strategy |
| 01:52:15 | <anarcat> | it would be the first layer, a "zfs send" that would replicate data to a secondary site, which then actually becomes a hot spare for the service as a bonus |
| 01:52:33 | <ivan> | I store my local data on btrfs with one filesystem per disk |
| 01:52:35 | <anarcat> | it can do its own periodic snapshots to keep away ransomware attacks and so on |
| 01:52:49 | <anarcat> | then you have something that replicates *that* to another, offsite server, with something *else* than zfs |
| 01:52:54 | <anarcat> | that's the part i'm not quite sure of yet |
| 01:53:10 | <anarcat> | borg or restic, maybe |
| 01:53:20 | <ivan> | Internet Archive stores data on one ext4 per disk (filled up to 100%) and tracks it in mysql (I think?) |
| 01:53:49 | <@JAA> | Yep, borg and restic is what I was going to mention. :-) |
| 01:56:58 | <hexa-> | bupstash seems to be a new kid on the block |
| 01:57:07 | <hexa-> | https://github.com/andrewchambers/bupstash |
| 01:59:37 | <anarcat> | it's "beta" |
| 02:00:25 | <hexa-> | yeah, for "REDUNDANT" backups :D |
| 02:00:32 | <hexa-> | all new backup software is beta :) |
| 02:04:32 | <anarcat> | i don't have "redundant" backups |
| 02:04:39 | <anarcat> | you have backups that work, or that don't |
| 02:04:42 | <anarcat> | who does redundant backups |
| 02:05:00 | <@JAA> | Looks like an interesting project though. One to watch for the future, I guess. |
| 02:05:21 | <anarcat> | i'm not sure i see what it brings over borg or restic, tbh |
| 02:05:41 | <anarcat> | since attic came around, there's been a plethora of projects like this |
| 02:05:52 | <anarcat> | i find it quite confusing |
| 02:06:06 | <@JAA> | Offline decryption keys are a nice one that I think neither borg nor restic have. (I haven't read anything yet about how it works.) |
| 02:06:18 | <anarcat> | borg can have offline decryption keys, i think |
| 02:06:35 | <anarcat> | borg has two key modes, one with the keys on the client, the other on the server |
| 02:06:52 | <anarcat> | so you could probably implement that with borg |
| 02:07:00 | <anarcat> | although maybe you mean to not have them on the client as well |
| 02:07:10 | <@JAA> | Yes, not on the client either is what I mean. |
| 02:07:18 | <@JAA> | So that when a machine is compromised, it's not possible to read backups. |
| 02:07:42 | <anarcat> | okay this is actually interesting https://bupstash.io/doc/guides/Secure%20Offline%20Keys.html |
| 02:07:43 | <pabs> | anarcat, JAA: Oracle could release CDDL 2.0 saying "you can relicense CDDL stuff to AGPLv3+", that would be the best solution here (CDDL 1.0 allows them to do that) |
| 02:08:06 | <anarcat> | "Using a 'put key' lets you create backups without exposing your decryption key, while using a 'list key' let's the key rotate old backups based on queries, but without exposing the sensitive decryption key. " |
| 02:08:21 | <anarcat> | aka "don't trust the backup server" is something that's very hard (if not impossible) to do with borg |
| 02:08:26 | <anarcat> | pabs: true |
| 02:08:52 | <pabs> | well, maybe not AGPLv3+, thats Linux kernel GPLv2-only incompatible too :) |
| 02:09:48 | <anarcat> | ha |
| 02:09:52 | <@JAA> | Er, it's not really about trusting the backup server. That already only sees encrypted data and doesn't have access to the keys. |
| 02:11:09 | <anarcat> | sorry, i wasn't clear |
| 02:11:13 | <anarcat> | i meant "append-only backups"! |
| 02:11:21 | <anarcat> | where you don't trust the client, actually |
| 02:11:26 | <@JAA> | Yeah, that. |
| 02:11:32 | <anarcat> | it's the "ransomware" scenario |
| 02:12:04 | <@JAA> | You can do that with borg IIRC. |
| 02:12:44 | <@JAA> | https://borgbackup.readthedocs.io/en/stable/usage/notes.html#append-only-mode |
| 02:13:26 | <anarcat> | yeah so i think i might have written part of those docs, and they're a lie |
| 02:13:38 | <anarcat> | append-only mode works until you prune the backups, then they don't |
| 02:13:40 | <anarcat> | so you can't have both |
| 02:13:59 | <@JAA> | Well, yeah, but that makes sense. |
| 02:15:19 | <anarcat> | when you know the design, yes |
| 02:15:22 | <anarcat> | when you read the docs, no |
| 02:15:33 | <anarcat> | i think it's a major flaw in borg, personnally, and not one that's stated clearly enough |
| 02:15:36 | <@JAA> | Eh, it's kind of in the name. :-) |
| 02:15:55 | <Wayward> | I just keep a box of 2-1/2" USB harddrives with periodic offline backups and store them at my neighbor's house, and "working copies" on my own bookshelf or in use. A thumbdrive for weekly Documents/AppData backups. A harddrive for monthly system disk backup. I find it hard to appreciate NAS or cloud solutions after finding how frustrating they become. |
| 02:16:04 | <@JAA> | The entire point of 'append-only' is that you can only add to the backup repo, not delete *anything*. |
| 02:16:32 | <anarcat> | JAA: but then you might have the intuition that rotation is not deletion and that it's safe to prune *some* stuff right? |
| 02:16:43 | <anarcat> | in fact, you might even think that you could delegate that to the server instead, and that way you're safe |
| 02:16:46 | <anarcat> | but that doesn't work either |
| 02:17:26 | <@JAA> | Well, I wouldn't consider any destructive action safe. But yeah, delegating to the server would be nice. |
| 02:17:41 | <anarcat> | it doesn't work |
| 02:17:55 | <anarcat> | append-only works by marking destructive operations as pending and never executing them |
| 02:18:13 | <anarcat> | the second you act on a prune or whatever, even if only on the server side, the hostile client actions get applied and boom you lost |
| 02:20:19 | <@JAA> | I'm aware, and this is what I would expect. |
| 02:20:51 | <anarcat> | okay |
| 02:21:05 | <anarcat> | i didn't expect that, and only found out about it after setting up the service, so it was very frustrating :p |
| 02:21:30 | | sonick quits [Client Quit] |
| 02:25:18 | <@JAA> | Yeah, that would be annoying. |
| 02:25:45 | <@JAA> | How did you get out of that? |
| 03:46:02 | | Stiletto quits [Read error: Connection reset by peer] |
| 03:46:14 | | Stiletto joins |
| 04:04:00 | | ThreeHM quits [Ping timeout: 265 seconds] |
| 04:05:08 | | ThreeHM (ThreeHeadedMonkey) joins |
| 04:26:17 | | qwertyasdfuiopghjkl joins |
| 05:20:06 | | BlueMaxima quits [Client Quit] |
| 05:20:49 | <Wayward> | speaking of zfs backup server mishaps. oof... https://youtu.be/Npu7jkJk5nM |
| 05:37:26 | | AlsoHP_Archivist quits [Ping timeout: 240 seconds] |
| 07:08:26 | | Stiletto quits [Ping timeout: 240 seconds] |
| 07:25:06 | | Stiletto joins |
| 08:01:33 | | dvd quits [Read error: Connection reset by peer] |
| 09:01:48 | | sonick (sonick) joins |
| 12:55:58 | | Sluggs quits [Quit: Going offline, see ya! (www.adiirc.com)] |
| 13:02:46 | | Arcorann quits [Ping timeout: 240 seconds] |
| 13:05:40 | | sec^nd quits [Remote host closed the connection] |
| 13:06:49 | | sec^nd (second) joins |
| 13:07:41 | <yano> | https://worthless.pictures/ |
| 13:53:46 | | sonick quits [Client Quit] |
| 14:20:26 | <anarcat> | JAA: i did not, i still don't have a good story for this |
| 14:20:37 | <anarcat> | JAA: basically, i run offsites by hand |
| 14:41:08 | | sonick (sonick) joins |
| 14:59:47 | | AlsoHP_Archivist joins |
| 15:13:06 | | AlsoHP_Archivist quits [Ping timeout: 240 seconds] |
| 16:14:06 | | chrismeller quits [Ping timeout: 240 seconds] |
| 16:22:24 | | HP_Archivist (HP_Archivist) joins |
| 16:39:27 | | Sluggs joins |
| 16:48:01 | <@JAA> | Ah :-| |
| 18:18:40 | | AlsoHP_Archivist joins |
| 18:18:43 | | AlsoHP_Archivist quits [Remote host closed the connection] |
| 18:19:56 | | AlsoHP_Archivist joins |
| 18:20:13 | | AlsoHP_Archivist quits [Remote host closed the connection] |
| 18:21:26 | | HP_Archivist quits [Ping timeout: 240 seconds] |
| 18:21:26 | | AlsoHP_Archivist joins |
| 18:21:43 | | AlsoHP_Archivist quits [Remote host closed the connection] |
| 18:22:56 | | AlsoHP_Archivist joins |
| 18:23:13 | | AlsoHP_Archivist quits [Remote host closed the connection] |
| 18:23:39 | | AlsoHP_Archivist joins |
| 18:27:05 | | AlsoHP_Archivist quits [Read error: Connection reset by peer] |
| 18:56:27 | | sallysue joins |
| 18:57:41 | | Wayward quits [Ping timeout: 265 seconds] |
| 19:21:09 | | c41e3a (c41e3a) joins |
| 19:45:00 | | c41e3a quits [Client Quit] |
| 19:52:32 | | dvd (dvd) joins |
| 20:06:43 | | sallysue quits [Ping timeout: 244 seconds] |
| 20:22:22 | | c41e3a (c41e3a) joins |
| 20:22:30 | | c41e3a quits [Client Quit] |
| 20:49:10 | | dvd quits [Read error: Connection reset by peer] |
| 20:49:34 | | dvd (dvd) joins |
| 20:52:22 | | DogsRNice (Webuser299) joins |
| 20:56:15 | | xit78 quits [Ping timeout: 252 seconds] |
| 20:57:40 | | qwertyasdfuiopghjkl quits [Remote host closed the connection] |
| 21:09:11 | | sallysue joins |
| 21:14:08 | | qwertyasdfuiopghjkl joins |
| 21:26:39 | | qwertyasdfuiopghjkl quits [Remote host closed the connection] |
| 22:33:34 | | Arcorann (Arcorann) joins |
| 22:43:59 | | BlueMaxima joins |
| 22:45:22 | | dvd_ (dvd) joins |
| 22:49:41 | | dvd quits [Ping timeout: 265 seconds] |
| 22:50:10 | | dvd_ quits [Ping timeout: 265 seconds] |
| 23:44:18 | | Arcorann quits [Ping timeout: 265 seconds] |
| 23:46:55 | | Arcorann (Arcorann) joins |