[15:14:23] cdanis: any luck with the peering session? sorry if I missed any related update elsewhere [15:15:25] volans: it had already fixed itself [15:15:34] days ago [15:15:41] and there's no automated email on repair, merely on breakage [15:15:44] I decided to not reply to them because it would just open a ticket with them [15:15:54] nice! I agree [15:15:58] thx for looking [17:52:40] fwiw netbox-next is broken for a bit :) [22:25:26] is there a reason the netbox dns.git repo is really slow to clone? https://wikitech.wikimedia.org/wiki/Netbox#Exports I want to add it to codesearch, but don't want to cause any issues if it's expensive to serve or something [22:57:15] legoktm: interesting, that's new to me, but indeed is quite slow (I'm trying a new clone) [22:57:18] cc chaomodus ^^^ [22:57:37] Yah I dunno [22:57:47] if you could open a ticket about it i'll take a look [22:57:59] ok [22:58:19] if codesearch pulls it twice an hour would that cause any issues? [22:58:38] it's served by apache [22:58:44] (i think it's because it's a bare repository served over http ratherl than a real git server but I'll see what I can do about it) [22:58:50] i don't think it'll cause an issue [22:59:02] took 1m54.379s locally [22:59:25] while 0m3.957s to git pull [22:59:30] yah [22:59:32] legoktm: will it clone each time or just pull? [22:59:45] https://git-scm.com/book/en/v2/Git-on-the-Server-The-Protocols guess it's using dumb HTTP and not smart? [22:59:54] that's correct yes [22:59:59] volans: it clones with --depth=1 the first time and then just pulls [23:00:07] then should be no problem [23:00:14] I got [23:00:15] Fetching objects: 327, done. [23:00:18] 29 files changed, 80 insertions(+), 92 deletions(-) [23:00:22] real 0m3.957s [23:00:31] it was few days I wasn't updating it [23:01:47] legoktm: go for it [23:02:16] ty :) filed https://phabricator.wikimedia.org/T276403 [23:03:57] thanks :) [23:07:41] for context the reason why it's not smart is that we don't need/want RW capabilities,and seemed much safer this way that it can't be misconfigured [23:08:25] but is an interesting thing to keep an eye on and ideally fix if there is a nice solution [23:11:32] gotcha [23:27:41] Mar 03 23:27:22 codesearch6 docker[14068]: 2021/03/03 23:27:22 Failed to clone https://netbox-exports.wikimedia.org/dns.git, see output below [23:27:41] Mar 03 23:27:22 codesearch6 docker[14068]: Cloning into 'vcs-123f1d5b524e84f52a028c1f59e8b831b70f162e'... [23:27:41] Mar 03 23:27:22 codesearch6 docker[14068]: fatal: dumb http transport does not support shallow capabilities [23:27:45] I forgot about that :(( [23:29:17] there are only 356 commits so far, no branches [23:29:44] and yeah sorry I didn't saw the depth=1 above [23:32:03] I've run into this before, totally forgot about it [23:34:50] another option is to have Phabricator mirror the repository, and codesearch pull it from there [23:38:48] https://phabricator.wikimedia.org/source/netbox-exported-dns/ [23:48:00] legoktm: what would be the diff having phab mirror it? [23:48:09] I've replied to the task fwiw [23:48:16] phab supports shallow clones :) [23:48:51] but how does it replicates it? [23:49:03] how does it know there are new commits [23:49:18] also, is it RO? [23:49:26] polls based on a frequency of how often it gets commits [23:49:30] no, I made it just a mirror [23:49:52] https://secure.phabricator.com/book/phabricator/article/diffusion_updates/ [23:53:59] ack [23:54:11] https://phabricator.wikimedia.org/source/netbox-exported-dns/manage/basics/ currently says "15s" but I think that's just because it's still importing [23:55:42] anyway, time to bed for me... ttyl [23:56:17] night and thanks :) [23:57:17] https://codesearch.wmcloud.org/search/?q=registry2001&i=nope&files=&excludeFiles=&repos=