[00:16] wgrant: Could you supervise a txpkgupload rollout in 1h44m if I request it (and if we can find a webop)? [00:17] cjwatson: But of course. [00:17] cjwatson: We haven't deployed tmp-incoming yet, have we? [00:17] We have. [00:17] Oh. [00:17] RT#78156 [00:18] Oh, I mean the actual code. [00:18] I know the dirs exist. [00:18] Unless you mean we haven't deployed the txpkgupload that uses it yet, in which case that is correct. [00:18] But they're not yet used. [00:18] Right, so this is the first update deployment. [00:18] Yup. I got the deploymgr alias sorted out, so hopefully it's a two-liner. [00:21] The memory use on pepo hasn't changed in the last hour or two. I suspect it's correlated with script kiddies hammering on the ssh port, but hard to say. [00:24] Likely. [00:24] Thanks for fixing it. [00:25] ... we hope [00:45] Heh [00:59] cjwatson: Did you look at deleting most of txpkgupload.filesystem? That gets rid of most of the Zopey deps. [00:59] Most of the methods have been obsolete since poppy classic died. [01:00] I guess all the launchpadlib stuff is pulled in by oops-datedir-repo... [01:04] Not yet, no. Good idea. [01:04] In fact that entire interface should just be shot and rewritten. [01:05] Killing bzr from l.s's install_requires was some large percentage of txpkgupload's disk footprint all by itself ... [01:05] But that was just a quick win [01:08] Seems like it should be easy to remove from the test deps too. [01:08] The only reference is in a test about logging, and there's no bzr-specific logging code, so that could all end up in lp.codehosting. [01:10] Indeed it could. That was just more than two minutes. :-) [01:11] But it'd be a requirement for getting onto Python 3. [01:12] Yep.