=== xMopx is now known as xMopx1 === xMopx1 is now known as xMopx [06:21] good morning [06:42] I realized I like parentheses too much :-( given how often I wish there would be a general form of multi level [06:42] you know (what I [not really] mean)? :-) [06:44] so... using iostat i have a 12 Disk raid6. No underlying drive is using more than 50% but dm-2 (the lvm lv) is at 100%. is there a way to optimize this? [06:50] DirtyCajun: this isn't a problem [06:51] DirtyCajun: essentially it means "something is in flight" at any time [06:51] but it is not a "limit" as it would be with old style sequential queue disks [06:51] due to deep async queues on modern devices and raid on top this is not like 10 years ago where 100% means you can't go faster [06:52] cpaelzer, ok, if that is a non-issue, i then need to optimize the way that i am transferring the data to the disk to increase speed? [06:52] check the iostat man page on %util - it even says "But for devices serving requests in parallel, such as RAID arrays and modern SSDs, this number does not reflect their performance limits." [06:52] DirtyCajun: so your question converts to "how can I go faster" is that right? [06:53] yes. I am currently using bbcp so i can tweak threads unless i am hitting another bottleneck i dont see [06:53] box to box its going from a raid6 to a raid6 on a 10Gbps backplane [06:54] thoughts? [06:54] busy - back in a few mins DirtyCajun [06:55] ok [06:58] cpaelzer, actually... (for when you get back) according to a raid calculator my raid6 should only be getting about 180MB/s full write... so i assume im getting above that because im on the very begging of the disk [07:06] DirtyCajun: yes on spinning disks the beginning can be somewhat faster [07:07] just thew me off because of the %util per disk [07:07] im just impatient because its 48TB :P [07:07] so do you need/want to be faster - or did you just wonder about the 100%? [07:09] the 50% on the actual disks seem to indicate that you could go faster if you push faster [07:09] but performance always depends on details, so take that sentence above with a grain of salt [07:09] s/push faster/push harder (i.e. do more i/O) [07:10] Good morning [07:10] hi lordievader [07:10] Hey cpaelzer [07:10] How are you doing this monday morning? [07:13] lordievader: a bit nervous for non IRC comatible reasons, but other than that good :-) [07:13] and how is your week starting lordievader? [07:13] DirtyCajun: I'm no bbcp expert, but more streams could do what you need [07:13] Like normal. I suppose that is a good thing 😃 [07:13] (to push harder and therby go faster) [07:14] yeah normal=good lordievader [07:14] i bumped it from 4 streams to 8 and its literally didnt change anything [07:14] Thanks for the confirmation :) [07:14] it think its gonna be what its gonna be [07:15] DirtyCajun: if you are ok with it I'm fine :-) [07:15] and its all over the board. its just the average is about what i said above [07:15] http://termbin.com/zc63 [07:16] crazy how it can hit 53 as a low and 550 as a high in less than 30 sec [07:17] DirtyCajun: depending how it copies I/O block sizes are very important [07:17] i have it set at 4M blocksizes [07:17] so if it copied files as-is a bunch of small files might be slow and huge one fast - depending on how it writes to the target disk [07:17] none of these files are <10GB [07:18] well that qualifies as non-small :-) [07:18] so it SHOULD be sequential read and write [07:18] hahaha [07:18] its more all over the place than my wife [07:18] last one went at 439MB/s [07:19] then it will just crap the bed for a few seconds. Im assuming something to do with write cache/ dumping from write cache during that itme [07:19] time [07:19] you might be hunting just a bad average calculation algorithm [07:19] thats true. [07:19] maybe if you'd report every 30 seconds you never would have spent time to thnk about it [07:19] yeah. im looking at the 1 sec differences not the average on the right. fair. [07:20] and it is documented to scale the tcp window sizes [07:20] so if it hits a max and throttles back with so much in flight you might veen have such a short drop before you are back up [07:21] i set it to 60 second polling and im gonna go chill out for a bit. Its pumping through to grafana so i can look at it that way as an overall stat [07:21] thanks for the time cpaelzer [07:28] yw DirtyCajun [09:06] good morning [09:07] I am following the story of this bug for Ubuntu Xenial https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1738219 [09:07] Launchpad bug 1738219 in linux (Ubuntu Bionic) "the kernel is blackholing IPv6 packets to linkdown nexthops" [Medium,In progress] [09:07] Fix Commit is the last step before Fix Released ? [09:08] I mean, packages should be available in proposed, right ? [09:11] I added http://ch.archive.ubuntu.com/ubuntu xenial-proposed/main amd64 Packages to my source list [09:12] but how do I understand which kernel version has the fix for that bug ? [09:12] I did [09:12] apt changelog linux-image-generic [09:12] but I cannot find the bug in the changelog === led2 is now known as led1 [14:10] jamespage: would be nice if you could target this for bionic https://bugs.launchpad.net/ubuntu/+source/ceph/+bug/1750376 [14:10] Launchpad bug 1750376 in ceph (Ubuntu) "ceph-volume binary missing from 12.2.2 packages" [Undecided,New] [14:22] frickler: yes of course [14:23] frickler: btw ceph-disk is not deprecated as of 12.2.2 (that got reverted) [14:23] it will be as of next release tho [14:23] with removal the one after [14:24] jamespage: well, http://docs.ceph.com/docs/master/ceph-volume/#migrating still claims so [14:24] jamespage: but yeah, ceph folks are doing lots of (not so) funny things currently [14:26] I was going on the ML thread from pre-christmas [14:31] rbasak: hi, I'm doing a dist-upgrade on a xenial server, and this was shown: [14:32] The following packages will be REMOVED: [14:32] python-certbot [14:32] rbasak: did that package change names? [14:32] hm, python3-certbot is a NEW install [14:32] I have python-certbot 0.19.0-1+ubuntu16.04.1+certbot+1 installed [14:33] apt-cache policy doesn't show from where, just the local dpkg db [14:33] maybe it was a package from upstream? [14:33] fginther: hi, do you remember where you installed that python-certbot package from? It's landscape's jenkins [14:33] oh, holiday probably [14:35] ahasenack: that doesn't sound like a package from the archive [14:35] oh, you are right [14:35] python3-certbot comes from a ppa [15:37] ahasenack: could you see if https://jenkins.ubuntu.com/server/job/git-ubuntu-ci/288/rebuild works for you please? It just seems to redirect me back without starting a rebuild. [17:02] rbasak: checking that url [17:04] rbasak: I see a build, and I'm not logged in. No button to trigger a rebuild. let me login now [17:04] rbasak: now I see the "rebuild" link on the left menu [17:04] Ah [17:04] I appear to be logged out. [17:04] Sorry! [17:04] Thank you for looking for me. [17:04] I should have thought to check that. [17:05] it's definitely not obvious [17:17] rbasak: thanks for the ua-tools review. Are you tagging and uploading? [17:17] ahasenack: already done :) [17:17] nice, thanks [17:18] I'll keep an eye on excuses [17:24] i have a folder with chmod 2770 to which i'd like to mount a network share. https://wiki.ubuntu.com/MountWindowsSharesPermanently does not tell me how i could recreate the permissions on the folder, where should i read up on that? [17:32] j4g0: are you sharing between two linux boxes? Or the server is windows and the client is linux? Where is this chmod? [17:33] server is a kazooli sme server 9.2, which is based on centOS, the box i want to mount on is ubuntu 16.04. The folder i have is on my ubuntu box. [17:35] i only found a description on how to set the gid in the fstab file on the mounting box but that only means the folder is set to chgrp setting a gid is not the same as permissions (2770) [17:36] true, hence my question on how i could recreate those permissions [17:36] what is it you want to do, force all files to be owned by a certain group? [17:37] yes, i want all files in said folder to belong to a certain grp [17:37] the server part, if that's samba (most likely it is) has options for that, per share [17:37] ah ok, tyvm [17:37] then i know where to look that up [17:37] check the smb.conf manpage, look for the "force group" parameter [17:38] there are other force parameters [17:39] that won't change already existing files [17:39] if you need those to be owned by a specific group on the client side, then I think you need to look at mount.cifs options [17:39] specifically, forcegid [17:40] but if that gid belongs to a different group on your client then it does on the server, you will see different names when listing the files [17:40] different group owners, I mean [17:41] i'll try the mount.cifs first then, tyvm === Guest92838 is now known as devil__ [20:17] Hey, I am having an issue where the right shift key will not work while connected over xrdp. The left one works ok. I have tried the suggested steps checking the layouts in /etc/xrdp. The Keycodes match for 50 and 62 correctly. I have tried to map the right shift key to the left action also as a last ditch action but this just makes many other keys not work lol. [20:19] ? [20:19] question mark for me? :)