| jvelasquez | hi. I made /etc a repo, added everyting in /etc to it, commited, and then wanted to "relocate" (terminology from svn) the repo to a real server, so I did a 'switch --force', but now I have nothing in /etc. | 19:54 |
|---|---|---|
| jvelasquez | is there an easy way to restore my /etc? | 19:54 |
| jvelasquez | or any way? | 19:54 |
| jvelasquez | I suppose it's on my server and I just need to check it out. :| | 20:04 |
| jvelasquez | damn. there's nothing on the server. | 20:19 |
| jvelasquez | I wonder what this command did: bzr switch --force bzr+ssh://j@jv.lan/srv/bzr-etc/trunk | 20:20 |
| mgz | jvelasquez: pastebin `bzr info -v /etc`? | 20:26 |
| mgz | you should still have the repo under there, but you've pretty seriously borked the branch | 20:26 |
| jvelasquez | says: bzr: ERROR: Unrecognised value for BZR_SSH environment variable: ssh | 20:30 |
| jvelasquez | I think the data is gone. /etc/.bzr doesn't have much of anything. | 20:31 |
| jvelasquez | I guess it's gone. I'll start rebuilding from new. :| | 20:32 |
| fullermd | I wouldn't expect it to really blast away the repo... | 20:35 |
| fullermd | Though I couldn't even guess what switch --force would do on an initially standalone branch. | 20:36 |
| fullermd | There might be micro black holes involved. | 20:36 |
| jvelasquez | well, what was I supposed to do? | 20:37 |
| jvelasquez | I hope, at least I can learn the right commands. | 20:37 |
| fullermd | Mmm, probably depends on a lot of details, but my first impulse would be something like push ; bind. | 20:37 |
| fullermd | How'd you populate the j@jv.lan/srv/bzr-etc/trunk branch before the command? | 20:38 |
| jvelasquez | opened a shell on that machine directly, and did, bzr init-repo bzr-etc; cd bzr-etc; bzr init trunk | 20:39 |
| fullermd | 'k. So you created an empty branch, then tried to switch your existing /etc over to it. So, winding up with an empty tree makes sense. | 20:39 |
| fullermd | I would _presume_ that there's still a bunch of stuff in /etc/.bzr/repository though. | 20:39 |
| fullermd | (I mean, it'll probably be small, since /etc is all little tiny text files) | 20:40 |
| fullermd | So your WT would be screwed up, and so would the branch. But theoretically, the repo is fine. | 20:40 |
| fullermd | So... | 20:41 |
| * fullermd contemplates. | 20:41 | |
| jvelasquez | yes. lot's of stuff in /etc/.bzr/repository | 20:41 |
| fullermd | Well, first off, tar up /etc/.bzr just as a backup, so when I suggest something stupid you only get annoyed instead of hunting me down. | 20:41 |
| fullermd | Then... hm. OK, copy/untar the whole .bzr in /tmp/etc (or whatever scratch space) | 20:42 |
| fullermd | Manually blow away the checkout/ and branch/ dirs inside .bzr. | 20:42 |
| fullermd | Hm. Probably can't just use 'init' blindly, but... | 20:43 |
| fullermd | 'bzr reconfigure --use-shared' should make the repo think of itself as shared, then you can 'bzr init xyz' to create a /tmp/etc/xyz empty branch. | 20:44 |
| fullermd | Then from in there, you can use 'bzr heads --dead' (from bzrtools, I think) to fish out the old head revision id from the repo, and then use 'bzr pull --force -rXYZ .' to pull it out. | 20:44 |
| fullermd | And check it over. If that works, you can just do a pull from that into your switch-ified /etc to blat the info up into the new repo and restore the WT. | 20:45 |
| fullermd | (won't restore permissions though, so you'll have to do that manually, or master.passwd or shadow or whatever will be world readable, etc) | 20:45 |
| jvelasquez | etckeeper was also setup, for permissions | 20:46 |
| fullermd | That'd make that part easier. | 20:46 |
| mgz | fullermd: thanks for doing the brain-work | 20:49 |
| fullermd | Well, it was either that or do my _actual_ work, and that's boring and annoying ;) | 20:50 |
| jvelasquez | yes. thank you. I've been too busy following your steps, than to thank sooner. | 20:55 |
| fullermd | Well, don't thank me 'till it works :) | 20:56 |
| jvelasquez | before `bzr reconfigure --use-shared`, /etc is 12MB. after is 9MB. it asks me for my passwd to the other host, twice. | 21:05 |
| fullermd | Mmm. I was saying to do that off in a temp copy, after blowing away the branch/checkout dirs; it shouldn't be trying to contact anything else then, just flipping the bit in the repo. | 21:07 |
| fullermd | Maybe it did a repack or something. | 21:07 |
| jvelasquez | ahh! "blowing away the branch/checkout dirs"! | 21:07 |
| jvelasquez | I'll repeat. | 21:08 |
| jvelasquez | your right. it doesn't ask any more. but it also dumped it's repository. | 21:09 |
| jvelasquez | maybe I can return the repository without it noticing. | 21:10 |
| fullermd | Hm. | 21:17 |
| fullermd | I thought reconfig would alter the repo. I mean, it tries doing stuff with the branch too, but I thought on a bare repo... | 21:18 |
| fullermd | Wow, it completely blows it away. | 21:19 |
| jvelasquez | I think I got it! | 21:19 |
| jvelasquez | yep! | 21:19 |
| fullermd | Goodie, I was looking for an excuse to file some bugs to harass mgz... | 21:19 |
| jvelasquez | basically exactly what you said. | 21:19 |
| fullermd | 'k, I think the 'bit' is just a file. | 21:20 |
| fullermd | So touch .bzr/repository/shared-storage in place of the reconfigure step. | 21:20 |
| mgz | fullermd: :P | 21:20 |
| jvelasquez | I think I'm actually starting to learn some of this. | 21:21 |
| jvelasquez | fullermd, thanks so much. maybe after friday, I can send you $50 | 21:21 |
| jvelasquez | least i could do. :| | 21:22 |
| fullermd | Oh, that's unnecessary. | 21:22 |
| fullermd | 'course, after mgz notices me reassigning all my bzr bugs over to brz, I may need the getaway money... | 21:23 |
| jvelasquez | and so is the effort to help. | 21:23 |
| fullermd | Eh, I consume support, I provide support... it all evens out in the end. | 21:26 |
| fullermd | As long as nobody tallies my consumption too carefully, anyway. | 21:26 |
| * mgz chews on fulldermd | 21:27 | |
| mgz | -d | 21:27 |
| mgz | too much of a mouthfull | 21:27 |
| fullermd | 4 out of 5 dentists advise against recreational fullermd chewing. | 21:27 |
| mgz | only because they haven't tried it yet... | 21:30 |
| jvelasquez | ok. so now for a "push" and a "bind" | 21:30 |
| fullermd | 'push' will take what you've got in that local branch and make a copy of it in the remote location, and then 'bind' ties your local branch to that remote so commits automatically go upstream. | 21:36 |
| fullermd | Effectively moving you from "local branch owns everything" to "remote branch owns everything, and i have a local checkout" | 21:36 |
| fullermd | (modulo my usual checkouts-vs-bound grumblings, anyway) | 21:37 |
| jvelasquez | has anyone ever seen a bunch of your files, all renamed with the postfix str ".~1~" | 23:16 |
| fullermd | That's where things get renamed when bzr needs to move them out of the way. e.g., 'revert'd changes, pre-existing unknown files when you merge/pull in new files with conflicting names, etc. | 23:22 |
| jvelasquez | `bzr info` lists Related branches, but the push branch is wrong. I'd like to update it. | 23:26 |
| fullermd | push --remember | 23:27 |
| jvelasquez | :) | 23:28 |
| jvelasquez | to think i trashed the whole working copy and repo, just cause of that old push branch. | 23:28 |
| jvelasquez | So, I did a push to a remove location, inside of a repo, inside a shared repo, but yet when I open a shell local to the repo, it says the content I added is "Unversioned" | 23:41 |
| jvelasquez | I guess it's versioned here, locally, and it's been pushed to a remote location, and that remote location is inside a repo, but the files were not actually added to the repo. | 23:42 |
| fullermd | I suspect you're probably mixing up what's where with working trees, etc. | 23:43 |
| fullermd | Particularly, pushing to a remote repo won't touch anything in its WT. And as a central repo location, there's no reason for it to _have_ a WT in the first place. | 23:44 |
| jvelasquez | WT = working tree? | 23:45 |
| * fullermd nods. | 23:45 | |
| jvelasquez | ok. so do I have it right? | 23:46 |
| jvelasquez | I want to version my files, and centralize their location. they say "unversioned", and this is ok! | 23:49 |
| fullermd | It depends on exactly what "they" are in this context. | 23:49 |
| fullermd | If "they" are the files in the location where you're doing stuff (i.e., somesystem:/etc), no. | 23:50 |
| fullermd | If "they" are files sitting aroud in the filesystem where your central repo is (i.e., jv.lan:/src/bzr/etc/trunk), who knows; that has no necessary relation at all to what's in the _bzr repo_. | 23:51 |
| fullermd | (which is why I say, generally don't even have WT's in central repo places; looking at them will only ever confuse things) | 23:51 |
| jvelasquez | I guess they are not being edited. They're being centralized, so that I can check them out from other hosts, where they will be Hot or Live | 23:52 |
| fullermd | The simplest way to be sure of what's actually in that centralized branch is probably just to make a scratch checkout in /tmp or somewhere, and make sure you get what you expect. | 23:53 |
| jvelasquez | ahh. so using term "trunk" is inappropriate here | 23:53 |
| fullermd | It's a social term, not a technical one; it's a matter of how you use and arrange your pieces. The pieces themselves just sorta float around independently. | 23:55 |
| jvelasquez | but I believe that's what I want. cause I'm collecting a few repos which are "current effort" aka trunk, but later I will collect other similar sets of repos, which are different branches from it. | 23:55 |
| fullermd | Sorta like "central repository" (which probably more strictly means "central branch" in this case); there's nothing technical intra-bzr you do to make it Central, what makes it Central is just how you use it. | 23:56 |
| jvelasquez | ok. but the versioning was originally done on my machine, and then I just pushed the whole repo to a fileserver for later. | 23:57 |
| fullermd | The terminology can trip things up a bit; technically you pushed the branch. | 23:58 |
| fullermd | You pretty much never interact with a _repository_, only with a branch. The distinction in speech often doesn't matter, but sometimes it can cause confusion when discussing details. | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!