[00:00] pstate is an Intel alternative for cpufreq, but it's possible that it doesn't work well with all CPUs, I don't know... [00:01] in theory it shouldn't affect # of CPUs though [00:01] and the two nomdm* are about RAID [00:03] I have to ask those guys at server shop then [00:04] or test box with another OS [00:04] I suspect they will not be able to help you :) [00:04] other OS seems like a reasonable enough test, but another *cpu* frmo a different supplier might be more useful still [00:04] I have 5 year warranty on such hardware... [00:05] might be faulty cpu microcode too ? [00:06] got error on it last time I ran apt upgrade [00:08] well at least that one might be easy to nail down, uninstall the intel-microcode package, rebuild initrd, reboot, re-test, and check the microcode date in dmesg.. [00:08] E: intel-microcode: failed to create or prepend the early initramfs to the initramfs [00:08] it seems insanely unlikely to me though [00:09] we'll publish a new microcode in a few days for the spectre v4, maybe your cpu is covered in that list... [00:09] -> /usr/sbin/iucode_tool: microcode bundle /lib/firmware/intel-ucode/list: unknown microcode format [00:10] downloaded from intel: microcode-20180425.tgz [00:21] Blueking: my guess would be that it's actually a BIOS/UEFI firmware bug (possibly in ACPI) [00:22] hmmmm.. I wonder if the fwts package could help? [00:24] I wonder if there is a kernel parameter to ignore ACPI's core count (or whatever makes this wrong) === uplime is now known as klime === klime is now known as uplime [05:12] good morning [05:57] Good morning === jaskal_ is now known as Guest1700 [08:54] sarnold, I do not have access to Better ARK but people on #clearlinux irc channel do [08:54] sarnold, lack of CPUs may indicate custom microcode that disables/enables them; or that cpus were fused away, meaning there is a crippled SKU out there aka a cheap edition. [08:55] sarnold, many of Intel CPUs are the same, and artificially clocked down / cores disabled. And one can sometimes pay to unlock higher clock speeds & cores. === jaskal__ is now known as jaskal_ === jaskal__ is now known as jaskal_ [09:22] ahasenack: I'd say the reporter has modified his system from defaults so much in bug 1772337 that it's likely to be a configuration problem than a bug in Ubuntu. If he thinks it's a bug he should at least present an explanation as to why. So Incomplete. [09:22] bug 1772337 in mysql-5.7 (Ubuntu) "package mysql-server-5.7 5.7.22-0ubuntu18.04.1 failed to install/upgrade: installed mysql-server-5.7 package post-installation script subprocess returned error exit status 1" [Undecided,Triaged] https://launchpad.net/bugs/1772337 [13:15] can some one please tell on 18.04 i have process with sha256sum and it keep running for long time and it kepp cpu busy ,how do i find what it is doing and trace it ,and it keep changeing pid even kill within second it start new same process ,is it normal ? [13:17] samba35: try ps with the f option to see hierarchy [13:17] ok ,i will try it when this start [13:17] did you put some security stuff who want to have hash from each file ? [13:18] how do i find which package hold this file /usr/bin/sha256sum [13:18] apt search sha256sum ? [13:19] dpkg -S /usr/bin/sha256sum [13:19] hm, mod-wsgi dep8 tests passed locally [13:19] * ahasenack clicks the retry button in excuses [13:19] ps auxf should be ok ? [13:19] exactly how I use ps :) [13:20] but someones prefer ps -edf [13:20] thanks Ool and ivoks [13:21] ok will that too :) but thanks for give idea get good idea [13:24] my shell script functionality, what would be good else to add there? [13:24] https://paste.ubuntu.com/p/ChMScTw5sp/ [13:24] backup, generation and set up sshkey, set up certificates for virtualdomains [13:25] I think about create a fullfladgit control panel :) [13:52] Hello everyone, basic question but can not find answer easily on my side. Let's assume I have a dhcp server on my network which assigns IP AND hostname. When my machine boots, it gets the IP but hostname command and /etc/hostname file contains ubuntu. Machine is ubuntu xenial 64bit with minimal set of packages. Furthermore my seconday IP address is on another network served by another dhcp server but this one does not provide host [13:52] names in DHCP offers. How does all this works and what should I add/change to get my hostname properly set ? [13:54] I don't know [14:02] olivierb-: if you're using ifupdown on Xenial, then it's done via dhclient.conf and dhclient-script [14:02] I'm not sure what the default is or if you need to change it. [14:03] olivierb-: it's probably not a good idea to have a server's hostname regularly change though. Unless it's an appliance, I'd lock it in statically on the client side anyway. [14:03] rbasak I managed to get it working once using service networking restart, but it was only once [14:04] name should not change as long as the lease remains same [14:18] there is this fence-agents package in the ubuntu archive, that ships a lot of scripts of the type fence_* [14:18] there is one of those scripts that doesn't work out of the box because of missing python dependencies [14:18] these missing deps are declared as "Suggests" in the d/control file [14:19] that seems wrong to me. If the script doesn't even run, surely those deps should be in Depends? [14:19] bug is https://bugs.launchpad.net/ubuntu/+source/fence-agents/+bug/1772096 [14:19] Launchpad bug 1772096 in fence-agents (Ubuntu) "fence_vmware_soap is not working after installation of fence-agents" [Undecided,Triaged] [14:19] check comment #2 [14:19] and #4 [14:19] it's a sync from debian, though [14:24] ahasenack: depends on the nature of it. AIUI, if it's not the normal use case of a package, it doesn't hae to be a Depends. [14:24] it's just one of many fence agents the package ships [14:24] take a look at apt-cache show fence-agents [14:25] "The Depends field should be used if the depended-on package is required for the depending package to provide a significant amount of functionality." [14:25] "The Recommends field should list packages that would be found together with this one in all but unusual installations." [14:26] "[Suggests] is used to declare that one package may be more useful with one or more others. Using this field tells the packaging system and the user that the listed packages are related to this one and can perhaps enhance its usefulness, but that installing this one without them is perfectly reasonable." [14:26] From https://www.debian.org/doc/debian-policy/#binary-dependencies-depends-recommends-suggests-enhances-pre-depends [14:27] it backtraces [14:28] For a package that is essentially a collection of a bunch of different use cases, it's difficult. [14:28] People might not want it to depend on every single possible use case dependency package. [14:28] they should have their own package each then, if it pulls in more deps than the maintainer is willing to accept [14:28] But by not doing that the user has to know what extra to pull in [14:29] the user will see a python backtrace [14:29] Yeah that's not good. It's not necessarily wrong to not have a dependency, but it's reasonable to expect a more helpful error message. [14:29] I would leave it to whatever the Debian maintainer thinks is appropriate [14:30] there's a changelog entry even, mentioning the two deps being added as build-depends, and suggests, at the same time [14:30] * Add python-requests and time to Build-Depends, add python-requests and [14:30] python-suds to Suggests (needed by fence_vmware_soap). [14:30] That sounds reasonable [14:32] I fail to see how [14:32] the package is shipping a binary, like many others it already has, that does not work because of a missing dependency [14:32] it crashes [14:32] apport will raise its head [14:33] It can be fixed by adjusting the apport hook or adjusting the error message when the "dependency" isn't found. [14:34] It doesn't necessarily have to be declared as a Depends unless it is needed to provide a significant amount of functionality for the package. [14:35] The decision is left to the maintainers otherwise, and it seems that they selected Suggests deliberately. [15:21] are there any dos tools that can post intel cpu, number of cores and so ? [15:27] you mean command line [15:28] https://www.tecmint.com/check-linux-cpu-information/ [16:06] Blueking: lscpu as before [16:33] rbasak: success! " * Move python3-requests and python3-suds to Depends (Closes: #899381)" [17:07] why usr/bin/sha256sum /etc/aide/aide.conf.d/31_aide_rngd keep cpu usage very hight [17:08] any idea why sha256sum is used for X11 relates file ? [17:08] because the CPU has to calculate the hash sum of a given file [17:08] why it's doing that I don't know, but sha256sum is not a 'light' function in terms of resources [17:08] (any hashsum function will eat CPU) [17:08] (to varying degrees depending on the filesize of the file being hashsummed, and the hashing algo) [17:09] yes but what kind of with X11 [17:09] I... have no idea what you just tried to ask. [17:09] which means either a language barrier, or your question was malformed./ [17:10] samba35: as I said, I can't answer *why* it's used for the X11 relates file, but I can answer the "keep cpu usage very high" part because of the nature of what sha256sum does [17:11] i once sha256summed a 240GB disk image file once. Took full CPU on the 16 cores of that workstation, for a good bit of time. [17:12] (100% usage, on pretty much all cores, until it finished the calculation of the sha256sum) [17:12] ok ,got it it seems im-config [17:12] remove this page and let see what happen [17:13] rbasak: little help: is this good? https://pastebin.ubuntu.com/p/ZPCZcYtykt/ [17:17] /etc/apache2/mods-available/proxy_http2.load now sha256sum is using this file [17:17] so its not static to one file [17:17] i am not trying encrpth /decrypth any file [17:18] it's not encryption/decryption [17:18] it's hashsumming. *something* on your system is trying to get hash sums for various files [17:18] samba35: you seem to be using aide. That will hash all files in your system [17:18] ^ that [17:19] how to i fix this ? [17:19] remove aide [17:19] or figure out why you installed it [17:19] unless this is not your system :) [17:20] is it relaed to check-security ? [17:21] don't know what that is [17:21] there is no such file or package in ubuntu bionic [17:21] brb will reboot system ,its keeping cpu very hight and its too hot already [17:22] rebooting won't fix it [17:22] it's likely a cron job, it will just run again [17:22] hmm how to open this ? intel-microcode_3.20180425.1~ubuntu0.16.04.1.tar.xz [17:22] tar won't work ? [17:23] tar xJf [17:23] J is for xz compression [17:27] hmm how do I implement microcode ? [17:27] none help file inside this package ? [17:28] I don't know, but there is a package in ubuntu that installs new microcode, you could check how it does it [17:29] are there search function inside apt ? [17:29] apt-cache search for packages [17:29] apt-file search for files in packages [17:32] Blueking: installing the intel-microcode package should be all you need [17:33] sdeziel how ? [17:33] Blueking: on your next reboot, a fresh microcode will be loaded if one is available [18:25] xnox: oh thanks for the #clearlinux tip -- I knew intel binned CPUs pretty heavily but I thought they were *really* diligent about making sure every unique combination had a descriptive name.. [18:25] HAHAHHAHAHAH [19:06] xnox: well not poor Blueking is likely to return his cpu because it's not what he thought he bought.. [19:08] sarnold, that is what ark.intel.com is for as there are too many to memorize [19:08] lyn||ian: ARK says Blueking's processor should have FOUR cores but it only has TWO cores. [19:12] wow' [19:13] yeah, that's what has me so surprised :) [19:14] I've spent so many hours reading ARK details on processors I've thought about buying.. [19:16] sarnold, well.... what does the box say? or the markings on the cpu itself? it might be sub-sku with disabled cores. [19:16] Blueking: hey :) do you still have the box? :D [19:42] sarnold it's a few months away from beeing used for 5 year :P [19:43] Blueking: so.. the box is long since discarded? heh [19:43] bios says 4 core [19:45] gotta find out how to remove current microcode and force new one in [19:48] apt-get install intel-microcode; make sure the initrd is rebuild; then reboot [19:48] Blueking: again, this should all be automatically done on reboot [19:48] Blueking: journalctl -k | grep -i microcode [19:48] but IIRC, this will only load a microcode on CPUs visible by the kernel... [19:49] last line didn't post anything [19:52] how to remove current microcode.. apt-get install intel-microcode says 'intel-microcode defined manual installation' [19:55] Blueking: you mean remove the intel-microcode package? `sudo apt-get remove intel-microcode` [19:56] ok used apt-get remove intel-microcode -> then apt-get install intel-microcode -> reply from process "/usr/sbin/iucode_tool: microcode bundle /lib/firmware/intel-ucode/list: unknown microcode format" and "E: intel-microcode: failed to create or prepend the early initramfs to the initramfs" [19:58] maybe this error make it fail update intel-microcode at next boot ? [19:58] Blueking: most probably [19:58] uh, don't think there should be a /lib/firmware/intel-ucode/list file [19:58] and yes, if the package failed to install, then it ... didn't install [19:59] Blueking: you asked how to untar the intel-microcode tarball earlier, have you extracted that tarball in that dir? [19:59] no [19:59] OK good [20:00] just did a quick chdist check and yea, there shouldn't be a list file [20:00] (at least, afaict) [20:02] I can confirm this /list file isn't here on my machines [20:02] maybe this was bad thing.. what I did 2 weeks ago ? mv intel-ucode /lib/firmware [20:02] tar -xvzf microcode-20180425.tgz [20:04] but only had two core before that [20:04] is that the tarball provided by Intel? [20:04] yes [20:05] I'd try to do the following: "apt-get purge intel-microcode" then "rm -rf /lib/firmware/intel-ucode" and finally "apt-get install intel-microcode" [20:06] ok [20:07] ok rebooting now [20:10] still 2 core [20:33] Blueking: did you have the "unknown microcode format" error prior to the reboot? [20:34] if no, I'd be curious to see the "journalctl -k| grep -i microcode" [21:08] sdeziel: http://paste.ubuntu.com/p/9JmBr77pvZ/ [21:13] sdeziel: interesting here http://paste.ubuntu.com/p/pSs9NcXDv3/ [21:14] enabledcores=1 looks suspicious [21:14] only one enabled core.. [21:14] you have HyperThreading right? [21:14] yes [21:17] OK so that explains the first paste [21:19] http://paste.ubuntu.com/p/M7nQ3bjjG5/ [21:21] so where do ubuntu have these control sets ? [21:22] grub/grub2 ? [21:24] sdeziel ? [21:25] Blueking: cat /proc/cmdline [21:25] sdeziel: http://paste.ubuntu.com/p/t6v8mqtqX2/ [21:46] Blueking: so there is no kernel arg that artificially limits the number of cores visible [21:54] there must be sumthin ? [21:54] anyway to check to make sure log rotation is working across web logs/etc.. ? or is this just on by default (16.04 server) [21:54] hello Janc [21:54] hello JanC [21:55] hello JanC_ === JanC_ is now known as JanC [21:56] JanC you had a peek at mine problem ? [21:57] Blueking: has this cpu always had two cores? :) [21:57] that I am not sure about [21:58] Blueking: yeah, I'd grep old boot logs for nr_cpu_ids [21:58] sarnold seems only one core are enabled with HT it's '2' core [21:58] cd .. [22:01] Blueking: currently my guess would be it's an ACPI problem? [22:02] http://paste.ubuntu.com/p/THJS4XjWH3/ [22:03] not sure [22:03] should I goin bios and change ? [22:04] what are RCU ? [22:04] RCU restricting CPUs from NR_CPUS=512 to nr_cpu_ids=2. [22:05] RCU is a go-fast lockless datastructure design scheme [22:05] it allows working with stale data in a predictable way [22:06] and it needs to know how many CPUs you actually have, I suppose [22:07] right [22:08] hmm I need to uncompress log files ? [22:08] to search through old log files ? [22:10] Blueking: zgrep should be able to look into .gz log files [22:10] zgrep -F nr_cpu_ids /var/log/syslog* [22:11] vim also has some autorules to expand compressed files in place [22:12] kern log files goes back only 3 days.. [22:14] Blueking: syslog, not kern.log [22:14] nothing there [22:15] do you ship logs to an aggregator/concentrator by any chance? [22:17] Blueking: I'm done for the day but good luck [22:20] gnight sdeziel ;) [22:20] thx [23:05] still a bug ? -> https://ubuntuforums.org/showthread.php?t=2187699&page=2 [23:08] Blueking: it's either still a *BIOS* bug (and will then surely remain that as BIOSes traditionally have a long lifetime) or it remains to be an ubuntu bug which, at least according to this forum thread, was not yet reported (and would thus not get fixed unless identified AND reported independently). [23:09] okey [23:12] tomreyn: configuration: cores=4 enabledcores=1 threads=8 with command lshw -class cpu [23:13] Blueking: so? [23:13] just wonder cores=4 enabled cores=1 but the number of threads=8 are it 8 enabled threads ? thus it indicates 4 enabled cores.. [23:14] just wonder if ubuntu read bios wrong where bios say 1 core someplace it means 1 physical cpu ? [23:15] it's supermicro mobo [23:15] Blueking: i don't think the threads relates to how many threads are enabled [23:15] Blueking: not sure, though, i'd need to read the source to be sure [23:16] Blueking: what does `grep -c '^processor' /proc/cpuinfo` say? [23:16] * tomreyn would also *assume* it refers to capabilities, not current configuration [23:16] inside bios there intel service thing wich tells status of mobo/cpu disabled cores = 0 [23:17] tomreyn: yeah, i assume givne the 'enabled' vs. no-prefix for cores, they are distinct for threads too (but possibly not emitting) [23:17] nacc nothing.. wrong typo ? [23:18] 2 [23:18] Blueking: ok, so you have 1 core enabled with HT on [23:18] Blueking: which is completely accurate with `lshw` [23:18] looks like that yes [23:18] Blueking: and your BIOS setting is currently set to 4 cores? [23:18] 'all' [23:19] active cores [23:19] Blueking: did you figure out what actual model (by the box) you have? [23:19] I can define 1,2,3, all [23:19] Blueking: based upon sarnold & xnox's disucssion above [23:19] Blueking: it seems most likely you have a buggy BIOS or you have a not-full-core chip [23:20] I *really* have trouble seeing supermicro having a bios this buggy [23:20] I know bioses suck and all [23:20] but defective-CPU sold at a discount is my current favourite theory [23:20] billing from shop: Intel Xeon Quad-Core E3-1230Lv3 1.8GHz 8MB, 25W, LGA1150 [23:22] shop didn't have this cpu on shelf.. were some ordering time and had to wait some time [23:23] nacc: I updated Bios 2 days ago to see if update fixed number of visible cores [23:24] I am gonna put another hdd on mine box and test with windows OS [23:30] sarnold: yeah, i would agree too [23:31] Blueking: also, fwiw, your mobo page didn't list this cpu as explicitly compatible [23:31] curious [23:32] I saw e3 v3-4 xeons on the page and though tit looked good enough [23:32] i saw e3 1200, and it wasn't immediately obvious to me if that included 1230 [23:34] this person claims to have seen 8 cpus in linux with a 1230: https://superuser.com/questions/851799/proc-cpuinfo-for-xeon-e3-processor [23:34] I think that's adifferent model, no L in that name [23:35] ah right [23:38] Blueking: this seems relevant: https://webcache.googleusercontent.com/search?q=cache:eWwSlZlUNDMJ:https://forum.pfsense.org/index.php%3Ftopic%3D128722.0+&cd=3&hl=en&ct=clnk&gl=us [23:39] nacc: DUDE [23:39] that looks like it exactly [23:40] they got an actual response from SM [23:40] i don't know if it's the same mobo, but the symptoms seem oddly similar [23:44] the only slight prblem of course is the bios upgrade was done to try to fix this issue [23:44] but still, clearing cmos and spending a few minutes to put it all back together again is quick and easy, easier than installing windows anyway :) [23:44] right