[04:00] rbasak, interesting.. thanks ! :) [07:09] Good morning [07:47] hi lordievader, how are you this week? [07:47] Doing okay so far. How are you cpaelzer ? [07:53] nacc: no it was not about the iscsi in xenial I know that we killed it there [07:53] nacc: instead it was the one in trusty by a new HWE form Xenial that exceeds two digit version numbers [07:53] lordievader: ok as well over here [07:54] 🙂 [10:08] HI all, I'm having issues with `mtr`, its only showing 7 hops, but tracepath is going to 30 before giving up [10:08] Any ideas why? [10:12] Takumo: some nodes in some paths block mtr's probes. [10:13] rbasak: does mtr not default to using the same ICMP probes as tracepath? [10:13] I'm not sure. [10:13] Looks like it uses ICMP echoes by default, but there are options to change that. [10:13] eg. --tcp [10:21] yeah, that gives me a whole bunch of different addresses for hop 6 , but still only 7 hops, 6 being the only one that replies [10:25] Sounds like you have stuff blocking stuff. [10:26] If tracepath does work, I'd start by running tcpdump/wireshark to see the difference and then try tweaking mtr settings to match. [10:28] ah, found it [10:28] TTL [14:05] Hi. Anyone here running Dovecot + Antispam + Spamassassin with pipe backend successfully? [14:07] I try running it on Ubuntu Server 16.04 and it looks like the plugin is doing exactly nothing. I can see its configuration being loaded with every IMAP interaction, but it is not doing anything. I added some logger and file touches in the wrapper script and also tried the spool2dir backend, which should write the mails-to-be-relearned to a folder. Nothing happens. No error messages and enabling debug just shows me that it is loading the [14:07] config. [14:07] Ideas? === jelly-home is now known as jelly [16:02] Hey guys. I've got a 14.04 LTS 64-bit desktop-edition server running in the cloud, running postfix and spamassassin. Recently started getting this error: /etc/cron.daily/spamassassin: [16:02] invoke-rc.d: initscript spamassassin, action "reload" failed. [16:02] Anyone running a similar configuration, seen this issue? Some googling showed me a few things, nothing really useful. I'm just checking here first, before I dive headfirst into the weeds on it. [16:19] cyphermox: Any further thoughts on: LP: #1732776? As far as testing, I'm tempted to change our automated tests to use priority to get things going. [16:19] Launchpad bug 1732776 in debconf (Ubuntu) "debconf/priority not respected" [Undecided,New] https://launchpad.net/bugs/1732776 === petevg_afk is now known as petevg [17:32] powersj: +1, priority will work for now and we'll fix the rest, clearly something was wrong seeing from my debugging session with infinity on friday evening [17:33] netcfg/get_hostname is also not followed [17:42] cyphermox: so this is a larger issue? [17:48] same issue for both, something is going wrong in the migration of the preseed keys from the kernel command-line to the installer's environment [17:49] so, I'm a bit surprised that priority alone works, but that might just be because multiple things go parse stuff in the command-line, and when you deal with aliases it gets a bit more complicated [18:00] Is there a ultra stripped down version of ubuntu server [18:00] one that is hardened and small kinda like alpine [18:42] I think you would also need a stripped down kernel etc. for that? [18:43] what would you strip? your 'needless module' is someone else's device in their boot chain :) [18:44] hm if landscape-client fails to register with a 404 error what did i do wrong? [18:44] i used the command it displays on the landscape computers page [18:55] hi, anybody doing something like ntop? I'm trying to figure out if and how I can setup a pcap ring buffer and generate some stats from it on the fly [18:55] for now analyzing 1Gbps [18:56] I think ntop does thing, with their PF_RING thingie [18:56] basically do a pcap capture to ram in a ringbuffer fashion and analyze it before it gets overwritten or something [18:57] the alternative, maybe simpler, would be to setup tpcudmp doing ring pcap writes and then have another process parse the last written pcap, maybe use a ramdisk for that [18:58] closest thing I could find is this, which actually kind of fit the bill: https://github.com/kumina/promacct [18:58] but my c++ sucks [18:59] the problem with pmacct is that it just does accounting, ie you can't do flow reconstruction and extract say a url from the pcap [19:00] which is also why netflow/sflow isn't a good chioce, altho I will prolly use that for a more lightweight accounting system with pmacct [19:08] drab: I'd look into netsniff-ng [19:08] nice, thanks, packet sniffing beast sounds good :) [19:09] speaking of stuff with horns... I also found this from bsd that is packaged for ubuntu: https://github.com/Cisco-Talos/Daemonlogger [19:09] basically a soft tap, which is nice [19:10] oooh zero-copy, nice [19:49] sarnold: they want their whole base system to be the size of what the Ubuntu kernel is now, so that's obviously not possible with a default Ubuntu kernel :) [19:49] JanC: ha :) [19:50] those are also often used as container-only distros, so many devices wouldn't be relevant [19:50] so then they just want lxd 'images' or something [19:50] you would have to ask that "vpndude" what they want [19:51] container/vm (not sure what they want) [19:51] * drab just built the smallest possible pxe bootable system for rescue [19:51] well, not true, but pretty small [19:52] I guess I could strip all the binaries and remove a bunch fo modules that i'm unlikely to use, but it's the smallest I could make while supporting all the modern hw [19:52] heck of a weekend, thanks to TJ- or I wouldn't have made it [19:53] just in case anybody is interested, because afaik there's no ubuntu based small system out there that works and can be pxebooted [19:54] drab: I assume the minimal install ISO can be PXE-booted :) [19:55] JanC: only to install afaics, not for rescue [19:55] ie it doesn't give you a full working system over pxe [19:56] it should be possible to drop to a shell IIRC, but it might not be very useful [19:56] some ppl are ok/happy with just getting busybox going that way and hitting alt-2, but that's not my case [19:56] right, that, not enough, I want an actual mini iso booted, basically a server system so that I can run a bunch of tools for diagnostic etc, kinda like UBCD [19:57] except that after trying UBCD kinda sucks ime and is not apt based/can't add tools at runtime like that [19:57] same for systemrescueCD [19:57] sort of like the rescue system some hosting providers have... [19:57] rootfs is 80MB, about 40MB of initrd [19:57] I guess, yeah [19:57] this is a full system tho, you can't tell the diff with say booting ubuntu server of an installed machine [19:57] it's no strippped down other than for what comes preinstalled with it [19:58] and the other requirements was to boot over http, withou the nfsroot stuff [19:58] which otherwise makes it possible to boot an install CD and "try before install" [19:59] the only thing I couldn't fix was dns in busybox... had to use ips in the pxe menu entry [19:59] e.g. https://wiki.hetzner.de/index.php/Hetzner_Rescue-System/en [19:59] but apart from that I'm pretty happy with the result and I have a whole new appreciation for ramdisks... with a local apt mirror I can convert into a full desktop if needed in 10sec