[00:02] I could really use a recent kernel with xen pv-ops [00:03] cj: if you click through to bug 418130, you'll see that we included an EC2 kernel with 9.10 [00:03] Launchpad bug 418130 in Ubuntu Karmic "Karmic EC2 images don't use a karmic kernel" [High,Fix released] https://launchpad.net/bugs/418130 [00:03] cj, but to answer your question, #ubuntu-kernel are the folks to talk to about it === dendrobates is now known as dendro-afk [00:13] mdz: ah, thanks === robbiew is now known as robbiew_ [01:02] hello to all. i have on ubuntu server a very strange problem with bind9. sometimes it works and sometimes it dont work. if i do at the moment a "nslookup www.wificom.ch" i get the error message "server fail !" accessing the site over the browser gives me the error message "ns.wificom.ch returned (NORECORDS)" [01:03] the strange thing is that bind9 worked with no probles last day but now i am asking me what the problem could be. [01:11] xperia2: Do you actually own that Swiss domain? [01:13] genii: yes ! it has worked also yesterday ! here is also the output of the "dig" command [01:13] http://pastebin.com/d65133b66 [01:13] the first part looks quiet good ! [01:14] the second part however refer to "67.215.66.132" [01:14] xperia2: whois is reporting correct IP of 80.254.182.249 [01:14] yes that is my fix ip to the server ! [01:14] ping works with no problem for the domain [01:15] just resolving of the domain dont work for some strange reason ! [01:16] xperia2: When did you make your last zone changes, and how? [01:17] genii: this are my last changes for bind http://pastebin.com/d65133b66 [01:17] Reading [01:18] if i want to run a graphical application on a ubuntu-server using X11 over SSH, does X11 need to be installed and running on the server? [01:19] genii: it looks like i have give you the wrong link ! this here is the right link with the bind9 zone changes http://pastebin.com/d4a8c82dc [01:20] for some strange reason "nslookup ns.wificom.ch" works and resolve good but all other domains like "nslookup www.wificom.ch" wont work [01:24] jjohansen, ping [01:24] could you look at https://bugs.launchpad.net/ubuntu/+source/udev/+bug/397187 [01:24] Launchpad bug 397187 in ubuntu-on-ec2 "[karmic] udev requires new kernel, breaks on EC2 (dup-of: 418130)" [Undecided,Confirmed] [01:24] Launchpad bug 418130 in Ubuntu Karmic "Karmic EC2 images don't use a karmic kernel" [High,Fix released] [01:24] sure [01:24] the final comments there. i think that CJ is saying that we dont have a paravirt driver available for karmic [01:27] hrmm, yeah I will take a look [01:27] xperia2: Looks like you may have some misplaced "@" there [01:28] yes i have removed the first mx line to the second bloch where also the maildomains for the other subdomains are declared [01:30] wanted to have separeted blocks for the domain and the mail resolving [01:30] will change it back in this case ! [01:36] xperia2: /etc/bind/named.conf.local has that box listed as it's master? [01:38] genii: yes http://pastebin.com/d45456a9 [01:38] xperia2: Ok. My dinner is ready, I'll return in 30-45 minutes [01:39] genii: okay ! have a nice dinner till later [01:41] ohh man this bind thing is really boring ! why does it dont work ! it is sure some stupid error. [01:42] Most often it's some typo in the file like a # instead of a ; for a comment line, or the yyyyMMddss isn't incremented each change, etc === chuck_ is now known as zul [01:53] genii: hmm whery strange ! have commented all mail entrys out and now it works ! [01:53] need to find what line of the mail entrys couses the problem ! [02:12] okay i think i have found the problem. [02:13] it works without any problems if i have only this line here [02:14] mydomain.com. IN MX 10 smtp.mydomain.com. [02:15] if i put a new line like this here for resolving mails for subdomains however it break bind9 [02:15] subdomain1.mydomain.com. IN MX 10 smtp.mydomain.com. [02:16] does anybody know how to resolve this problem ? [02:17] xperia2: What does: named-checkzone wificom.ch /etc/bind/db.wificom.ch report? [02:18] genii: if i do "sudo /etc/init.d/bind9 restart" it reports allways "OK" [02:18] xperia2: Thats not an answer to the question that i asked :) [02:19] i have assumed it :-) what exactly do i need to do for telling you that ? [02:20] xperia2: The named-checkzone will try to test your zone file and report any problems it can find. [02:20] So it may not report "OK" even if just starting up/restarting bind itself does [02:20] ahhh "named-checkzone" is a executable in this case ! okay give me a moment [02:21] okay at the moment it report [02:21] named-checkzone wificom.ch /etc/bind/db.wificom.ch [02:21] zone wificom.ch/IN: wificom.ch/MX 'smtp.wificom.ch' has no address records (A or AAAA) [02:21] zone wificom.ch/IN: loaded serial 2009111317 [02:21] OK [02:21] now i will try to comment out a subdomain mail line [02:22] So you need to declare smtp name before you can use it as something for other names to point to [02:24] ahhh true ! but is smtp really good for ubuntu-server postfix ? i have seen also other examples that use mail instead smtp [02:25] I don't think it really makes any difference [02:25] It's just a naming convention === chuck_ is now known as zul [02:26] ahh okay in this case i will stick with smtp as it is more clear for me [02:29] genii: great catch with named-checkzone [02:29] genii: named-checkzone wificom.ch /etc/bind/db.wificom.ch [02:29] dns_master_load: /etc/bind/db.wificom.ch:27: photos.wificom.ch: CNAME and other data [02:29] zone wificom.ch/IN: loading from master file /etc/bind/db.wificom.ch failed: CNAME and other data [02:30] xperia2: You might like to sometime check out the ubuntu-server guide, at http://doc.ubuntu.com/ubuntu/serverguide/C/ where DNS and other issues are covered at some length [02:31] xperia2: yeah - you can't have CNAME and anything else for the same label [02:31] genii: if i uncoment a a line for resolving the subdomain mail adress it produce a conflict with the allready existing subdomain [02:31] so i need to have for every subdomain a non CNAME entry hmmm [02:31] which is the best way to see all my installed packages to clean up if i can chose webmin or putty axx. [02:32] mail hosts shouldn't be CNAMES at all, btw [02:32] since they all get rewritten into the right hand side immediately [02:32] espacious: With dpkg --get-selections [02:32] dpkg --get-selections [02:32] got it. [02:32] thanks [02:32] dpkg -l [02:33] --get-selections will also tell you the packages you don't have installed [02:33] dpkg -l is better. [02:33] it has descriptions [02:33] genii: thanks for the docs link but as allways the documentations is not complete and it dont handle the things that i am working ! [02:34] espacious: Doesn't however tell you the installed/half-installed/deinstall states easily [02:34] i have allready looked at the configuration part for the dns part [02:34] xperia2: what made you think that you needed all those CNAMEs? [02:34] genii what command would? [02:35] lamont: as first i needed only subdomains and i have readed that it can be done very easy using cnames [02:35] (fwiw, CNAME basically says "anytime they ask for LHS, they really mean RHS", so having any other data makes no sense. On the bright side, bind9 made (or was it 8?) made that an error instead of just a warning that did the wrong thing [02:36] now i need the mail resolving for the subdomains and have started to put the new entrys [02:36] I fear that the doc you're reading isn't clear about the definition of "subdomain" [02:37] you only need MX RRs for whatever is on the right side of IN CNAME... if that needs to be different, then you don't get to be using CNAME [02:37] likewise, just because it doesn't have an A RR doesn't mean it can't get mail [02:37] didnt know that this will not work. Yeah exactly i have searched the web now for nearly about a week now but i can say you sure that subdomains arent documented very well. [02:37] mostly because the term is either meaningless or overloaded. [02:38] possible but it is used more and more with the time as the needing get bigger. [02:38] there are fully quallified domain names (FQDNs), which live in zones. zones contain one or more domains. [02:39] so, for example, archive.za.ubuntu.com lives in the ubuntu.com zone and has 'archive.za IN A ..." in the ubuntu.com zone [02:39] by traditional definitions, za.ubuntu.com would be a subdomain of ubuntu.com, but who cares? [02:39] lamont: okay i will do now the needed changes in my zone file and report back if it works ! [02:40] pulling resource records for za.ubuntu.com yields NXDOMAIN [02:40] meh. backwards on the example, fwiw [02:41] za.archive.ubuntu.com has an CNAME, archive.ubuntu.com has (more than one) A RR [02:41] zones always have SOA RRs and NS RRs or they don't work [02:42] lamont: is this line here okay ? [02:42] subdomain.mydomain.com IN A mydomain.com. [02:42] A RR takes an IP, not a name [02:43] what value does "subdomain" have? [02:43] in my case it is "photos" [02:44] and will the MX host for photos also be the MX host for mydomain.com? [02:44] for more to be exact [02:44] photos.wificom.ch IN A 80.254.182.249 [02:44] because "photos.mydomain.com IN CNAME mydomain.com" says "whenever they ask any question about photos, give them the answers for mydomain.com" [02:45] so if the MX list is the same, then CNAME will give you the MX list you want. if that answer is wrong, then it's not really "the same as", though it might have the same A RR [02:46] lamont: that with the MX Host is a litlle to much for my skills. i have postfix running on my ubuntuserver and i am able to send mails over a gateway [02:46] fwiw, "CNAME" ("Canonical Name") has to be the worst possible name for what it does [02:46] so... is wificom.ch also 80.254.182.249? [02:47] now i would like also to recieve the mails for all subdomains and maindomains on my ubuntuserver with posfix [02:47] and if I send mail to photos, does it go to that IP? [02:47] ah. [02:47] yes it goes all to this ip [02:47] note that if you have photos CNAME bar, then the mail will show up as 'user@bar', not 'user@photo'. This is the nature of email rewrites [02:48] which, if the users are supposed to be different, is not what you want. :-( [02:48] ah, so you're trying to move all mail handling to a diff IP than .249? [02:48] (and fwiw, your use of "subdomains" is what I would call 'hostnames"...) [02:49] lamont: wait wait ! i will post the right zone file for you [02:49] yeah - that should speed up the conversation [02:50] http://pastebin.com/d3f33dd47 [02:51] have now changed the cname entry for the subdomain as you have said [02:51] that's the whole file? [02:51] need only to make sure that resolving of the mail adress for the subdomain photos will works [02:52] lamont: at the moment yes will put with the time other subdomains [02:52] those times in the SOA are way long... if I ask you for a name, and you don't have it, my nameserver will cache that answer for a week, and I won't get the answer even if you add it in 3 minutes. (OT, but hey) [02:52] I avoid using a Neg cache ttl of anything over 3600, for anything other than localhost [02:53] so... in your file, "@" for a LHS (left hand side, first token) is == "wificom.ch", and any name that doesn't end with a '.' will have that appended to the name. [02:53] lamont: okay i will change that too. [02:53] so "photos.wificom.ch" could be written as "photos" [02:54] and line 22 could move up to between 15 and 16, and get an '@' [02:54] fwiw, dig requires a little bit more to understand, but is much more explicit in what it tells you than /usr/bin/host [02:55] okay give me just a moment to chnage this two things [02:57] * lamont wanders off for a couple min [02:58] lamont: sorry my laptop is for some reason heavy slow ! http://pastebin.com/d4f8820c [02:59] i have a lot of memory swap [03:00] no worries [03:01] lamont: are the changes done by me right as you have said or did i messed something up ? [03:01] so... when you've changed everything, and I send mail to user@www.wificom.ch, what IP should the mail be delivered to? [03:01] to the fix ip of the webserver where also the mail server and the dns server run and this is 80.254.182.249 [03:02] that is what will happen now. /etc/init.d/bind9 reload and then say 'dig mx www.wificom.ch' [03:02] normally i would however preffer to have the mail adress "user@wificom.ch" but i assume this is no problem [03:02] you'll get a CNAME RR, and then the MX rr for wificom.ch [03:03] okay will do ! [03:03] and the email will get rewritten into "user@wificom.ch" [03:03] that last bit being the somewhat sticky bit... [03:03] (depending on whether or not that's what you _want_ to happen) [03:05] lamont: it wont work as i have some syntax error [03:06] "near 'wificom.ch.': bad dotted quad" [03:06] the other thing about mail delivery: the rule is very simple. 1) fetch MX RRs for the target host. 2) if this host is in the list, remove every record with an equal or higher number in the priority field. 3) start with the lowest priority number and try them all until you run out. 4) if you still haven't delivered the mail, try A and AAAA RRs too. 5) try again later [03:07] its about this line here "photos.wificom.ch. IN MX 10 smtp.wificom.ch." [03:09] photos IN A wificom.ch. [03:09] wificom.ch. is not A.B.C.D [03:10] www.wificom.ch. 3600 IN CNAME wificom.ch. [03:10] wificom.ch. 3600 IN MX 10 smtp.wificom.ch. [03:11] those last 2 are what I get after I fix the photos line [03:11] okay think have found the problem ! [03:12] named-checkzone wificom.ch /etc/bind/db.wificom.ch [03:12] zone wificom.ch/IN: loaded serial 2009111325 [03:12] OK [03:12] http://pastebin.com/m50a96646 [03:13] that's after I munge a few things around, and drop the (now wrong) comment at the top, which also tells me that you started with the localhost file. [03:14] lamont: dig mx www.wificom.ch http://pastebin.com/d5c140ef2 [03:14] (a little reordering for clarity, and really pulling down the times in the SOA) [03:14] xperia2: that's what I'd expect [03:15] to better understand the dig output, see the output of: dig mx ubuntu.com. @ns1.canonical.com. [03:15] question is what you asked [03:16] answer is the answer [03:16] authority is the NS list for the zone that has the answers [03:16] additional is "other stuff you're probably about to ask for", and is not necessarily authoritative [03:17] lamont: okay will use your example zonefile! will now just dig mx ... [03:17] so when I'm reading dig output, I ignore missing stuff in additional, and get worried if it has wrong answers there. [03:17] /usr/bin/host has much prettier output, and is therefore much less precise [03:18] hear hear [03:18] I always prefer host for the first pass [03:18] twb: dig gets my attention, host gets maintained through the process of benevolent neglect. [03:18] and sometimes, I feel sad about that. [03:18] Or even getent hosts [03:20] xperia2: a bit more on those times (including the $TTL) - every answer your nameserver hands out comes with a "time to live" (TTL), which tells the requestor how long he may confidently cache the reply. That is, you are saying "here is the answer, and I promise that won't change for $TTL (seconds)" [03:20] promise accordingly. [03:20] lamont: okay i see now that the differnce is that ubuntu has AUTHORITY: 3 and i have AUTHORITY: 1 [03:21] ubuntu has 3 nameservers, you have one. [03:21] it differ also for ADDITIONAL: [03:21] Isn't it generally advisable to have a backup DNS? [03:21] ahh okay nice one :-) [03:22] twb: dont know where to put my other dns stuff ? some free dns resolver ? [03:22] additional is filled in as: "if I have the A RRs for any of the NS/MX hosts in answer, then additional, add them. If I don't, or run out of room, tough" [03:22] xperia2: therein lies the challenge [03:22] (finding a home for secondary) [03:22] zoneedit and dyndns [03:22] What could possibly go wrong? [03:23] twb: you trol., [03:23] * lamont relies on a network of friends-who-run-networks [03:23] More seriously, maybe hcoop (for personal stuff) [03:24] in the before time, one of the NS RRs for my home domain was palrel3.hp.com [03:24] or was it relay.hp.com. whatever, that's a decade old conversation [03:24] lamont: was that your workstation at work? ;-) [03:25] part of the NS RRset for 'hp.com'. I didn't admin it when my zone was added, but I was the one who dropped the zone as part of transitioning the administration of the machines to tier 2 support [03:26] may have been a diff host too, now that I think about it more. [03:27] lamont: if i compare the two dig outputs for me it looks like it should work as my dig output is the same as the one of ubuntu ! [03:27] and yes, I do know people who use dyndns and then update the host record in the registry so as to have the NS RR glue point to their machine. That way lies madness. [03:27] xperia2: right [03:27] lamont: there are little scripts to do that automatically [03:27] Erm, I meant to update the dyndns A record. Never mind, I'm dumb [03:27] twb: You mean dydns? [03:28] dydns == madness [03:28] ScottK: +999 [03:28] I was thinking ez-ipupdate and the other one that OpenWRT ships [03:28] well, the services out there.. dynamic zone updates (ala nsupdate) can be love. [03:28] They hook into dhclient or pppoe and do an https PUT or so at the end of it [03:30] lamont: i thank you a lot for your teaching and helping ! you are a great man ! :-) [03:30] no worries [03:30] the problem with DNS is that it really _is_ that simple. much of the confusion comes from people trying to understand it in the complexity they think it has [03:30] a last more general question before i quite. [03:31] i am planning since week to financial support ubuntu for theire great product especially the ubuntu-server [03:32] the problem is however the very high price in my eyes what they want for one year ! [03:32] i dont want to make them down or put the prices low as i know how hard is it to earn money [03:33] but from my experience the price for the ubuntu-server package is just a dream price and not a market price [03:33] dont know how much people buy this service packages but i itself would like to see some new service package [03:34] for private people like me that have a home server [03:34] I really don't have an answer to that - haven't ever looked into it [03:34] and dont use it in a company or for bussines [03:35] a good price would be about 40 to 60 Euros per Year or 50 to 75 USD [03:35] xperia2: For a home server user it is expensive. If you can't contribute in money, contribute in helping out. [03:35] xperia2: Generally home server users don't buy the support contract. [03:35] ScottK: I assume you'll be in dallas? [03:35] lamont: As long as this flu i currently have subsides, yes. [03:36] yeah - if it's like what I had, it sucks [03:36] yeah i prefer to spend money rather to help out as first i am not really that ubuntu geek and with the time i have really bad feelings about using ubuntu for having fun and knowing money could be needed for more success [03:37] and i dont think home users dont want to have a support package [03:37] this is something that ubuntu has to think about ! [03:37] xperia2: New people are the best ones for writing documentation for new people. [03:37] So there are contributions you can make. [03:38] i have lost more than a week for this dns stuff. with a home user service package it could be solved more faster and easy [03:39] and the best thing is people like lamont or other can earn additional money buy helping out ! i mean why does ubuntu dont provide this service to the public ? [03:41] xperia2: well, a fair chunk of my income derives from Canonical already... [03:42] from what i see and understand is that the ubuntu managment is loosing a lot of money becouse of this. lots of home users are using ubuntu and have problems. look only the ubuntuforums they are full of questions like how can i do this how can i make this ... [03:42] Why dont you provide for this people a service package and engage more people for solving this the professional way really ! [03:43] I haven't looked at the support packages Canonical offers (or others, for that matter), but I would think it reasonable to believe that the price is set in consideration of the costs of providing that service. [03:43] generally speaking, the open-source community has not embraced such a model. [03:44] lamont: i dont agree with you. first you need to ask yourself [03:44] administering a payment system for thousands of people helping others would likely not be cost effective for the company, especially when so many of us have no issue in doing it without charge [03:44] how much money would a home user spend for answering and solving his questions ! [03:44] xperia2: well, in my case? not much at all. [03:45] but then, I do have a bit of experience in s/w engineering and such. [03:45] i am speaking general about ubuntu as you have said the cost are high that why the price is high ! [03:45] (apt-cache show bind9 | grep Maintainer) [03:46] for me whit such a pilosophy you cant earn really money and that is why probably the open source is loosing money [03:47] xperia2: My experience is that most home users won't pay at all. [03:47] lamont: woow :-) [03:47] most of the folks I know doing support for fee are looking for clients with at least 12 computers, in an office, in order to consider taking them on as customers. [03:47] anything less than that costs them more money than they can possibly charge for it. [03:49] xperia2: I let others handle all the postfix questions though.... your bind9 struggles came across just as I was passing through, and CNAMEs are one of those not-understood things, so it seemed a good use of some time [03:49] My experience with small consulting jobs is that they just aren't worth the trouble (I've tried) [03:49] ScottK: +99 [03:49] mind you, for $100/hr, I'm willing to consider exceptions to the rule... :-) [03:50] somehow, most home users won't pay me that though. [03:51] esp just for the consult to see if I'll take them on. [03:51] that is, if I were doing that [03:51] I know somebody who asks 120 euro / h for programming jobs :P [03:51] lamont: this is a price that i will pay for a year :-) but look also how much does somebody cal you for problems in a year [03:52] xperia2: The problem is the expertise you want costs that much or more [03:52] xperia2: if I stepped up and answered every question like yours tonight, I'd be here 24x7 [03:53] i know that prices are high in the it world. but it looks like you dont want to understandme [03:53] ScottK: BTW, I got up to 40 euro for 2h from a home user (that I wanted to help for free), so don't generalise ;) [03:53] plus beer [03:54] I once got US$25 via paypal as a thanks, so I know it's not universally true. [03:54] Just not enough true to be worth the trouble [03:54] well, actually, this was for a non-profit, but he paid it himself [03:55] they run a local news agency as a group of volunteers [03:55] have 4 laptops running ubuntu, and sound didn't work, which wasn't funny as they provide local radio news ;-) [03:56] also, all local radio stations here run on Ubuntu + campcaster because thats' the only way they provide local news :-P [03:58] well okay your experince is sure also true on the other side as you have allready said [03:58] and also confirmed my version that [03:58] ubuntu is loosing good money by not providing a service package to there users [03:58] mostly becouse ubuntu will become problems with the costs as of the high prices ! [03:58] there are service packages, right? [03:58] xperia2: nothing stops anyone from offering ubuntu support for whatever they feel is a proper price. [03:59] Yeah exactly. Countrys that are unerdeveloped and dont have high living costs [03:59] xperia2: after being involved in running several companies over the past decade, I can tell you that there's a world of difference between answering questions in IRC when it suits my mood, and giving someone a phone number to call me (even just during business hours), and have to respond to their questions. [04:01] lamont: especially as often you have to investigate first too... [04:01] that you haven't found anyone offering support at the price point you want tells me that no one has decided it's worth it to offer that price point yet. [04:01] JanC: yep [04:02] lamont: i understand your position very good ! just wanted to answer t the fac if the costs are to high but the demand exist why not offering it by experts in countrys with less high personal costs [04:02] lamont: well, sometimes volume is a problem to offer lower cost too [04:02] JanC: big time [04:03] i mean in some countrys it people dont have even work and other earn maybe at maximim 300 USD per month [04:03] JanC: raise the price, the demand goes down... somewhere in there is a nice max in the income/week curve. [04:03] xperia2: and who are you saying should create this company to run support? [04:04] this are 3000 USD per year personal costs. i dotn understand really the ubuntu position why they dont offer a service package for private people ! [04:04] xperia2: there are countries where people earn < 10 USD / month too ;) [04:05] lamont: it should be sure ubuntu that organise that but the service could go over mail in english language like server-support-private@ubuntu.com [04:06] xperia2: every time I've seen a company _not_ offer a service, it's been because the projected income from the service did not justify the projected expenses to provide it. simple business. doesn't matter if you're selling soap, or computer support. [04:06] provided by sub companys or freelancer [04:06] lamont: that's not always true, there is also the need for money to do initial investments etc. [04:07] JanC: put together the proper business plan, and find funding. the cost of the money is just one more cost [04:07] making a business plan costs money too [04:08] and if you want to tie up millions in resources to get 10-20K net per year, you need to be a farmer.. because no sensible commericial lender will loan you that in general [04:08] ☺ [04:08] seems most farm loans are secured with land that the bank will sell to be developments if they ever foreclose. simple. [04:09] lamont: the cost arent the problem as you can start with people that earn 10 Euros per Month for answering the question. ! [04:09] well, the only bank in Belgium that saw a 30% increase in business last year was the farmers credit union ;-) [04:09] xperia2: after you vet them, and then there's the cost of administering the program. those would be the bulk of the expenses for it. [04:09] they are also one of the only banks that aren't publicly traded [04:10] what for a administration. i will goes all over mail ! [04:10] (they also own the largest internet bank & on-line trading website in Belgium though) [04:11] running everythin on linux ;) [04:11] private people write to a mail and explain her problems and wishes and the support answer it. thats way better than hanging for days in forums and waiting till somebody answer your question ! [04:11] xperia2: so lets say that I decided to start such a business. and lets say that you tell me you're a hot support guy and I should send you clients. how do I know that you are? how do I deal with the damage to my reputation for referring to you and you having unhappy clients? mitigating those risks all cost time, effort, and therefore money [04:11] xperia2: but professionals will also answer the easiest questions first [04:12] xperia2: and what you're suggesting is already done by the various support mailing lists and such, to quite an effective level once the requestor gets an understanding of how/where to ask [04:12] lamont: JanC: man this question is very easy ! [04:12] xperia2: trivial, actually. [04:12] which question? [04:12] JanC: his bind question [04:13] I didn't read that far back ;) [04:13] CNAME and other data [04:13] I once fixed someones IIS nameserver with the BIND documentation, so it can't be dificult ;-) [04:14] BIND issues are a bit like LDAP issues, in that the docs tell you how to manage the database, but the applications that use it are the ones that tell you what the data means [04:14] i am speaking about the question "how do I know that you hot support guy for my bussiness" [04:14] JanC: I prefer to fix IIS servers with a livecd [04:15] xperia2: ah, well that question - sure it's a simple one. It just requires time and effort to validate the answer in something beyond "trust me" [04:15] looks it is very easy: search for people in this countrys that have allready finished in the scool a it master or something like this [04:15] lamont: IIS DNS server works with BIND configuration files, but the IIS GUI fucks them up, so I told him to keep his hands away from the GUI ;) [04:15] well, that was years ago [04:15] heh [04:16] xperia2: such a search would not find me. [04:16] I doubt most peopel here have a masters ;) [04:17] lamont: well becouse you dont want. it would cost you maximal 5 minutes work to get your people with very high skills. do you want a example ? [04:17] xperia2: I hired into HPs MPE (OS) lab to work on device drivers prior to enrolling in a 4 year college. My studies were in electrical engineering. my paycheck was from software development [04:17] I'm not interested in starting such a business, so no... don't really need the example. [04:18] and I know plenty of people that have "proven" their technical abilities that I would never trust with supporting anyone. [04:19] If you want masters, go to #math [04:19] twb: slaves are better [04:19] wait. that came out wrong; [04:19] xperia2: looking back at your question, DNS is one of the more complicated things to understand [04:20] JanC: mostly because it doesn't actually use the data that it's storing, for the most part [04:20] JanC: isn't it basically just like NIS? [04:20] twb: only better, and different [04:20] Exactly [04:21] Maybe I'm weird in having grown up with NIS... [04:21] I remember when I was learning LDAP... my frustrations that none of the LDAP docs told me what I wanted... only to finally find that what I was actually looking for was documentation for the ldap-using app, not ldap [04:22] though the app's docs made little sense until I understood some LDAP, and back and forth we bootstrap [04:22] I don't have real experience with NIS, but even resolving basic DNS issues involves knowing what glue records & TTL are ;) [04:22] JanC: yep. [04:23] lamont: only a example for finding people that fit your needs for offering such a service. [04:23] and I don't pretend to be a DNS guru, but at least I know that ;) [04:23] go to the site it.com.mk visit the forum http://it.com.mk/forum/ go to the employment subforum [04:23] http://it.com.mk/forum/forumdisplay.php?f=136 [04:23] make a posting with the title "Ubuntu Specialist Wanted" leave your email and you are done. [04:23] though glue records can mostly be lalalaalalalalalalala for most basic questions that don't involve broken glue [04:23] BTW. Macedonia is that Land that have in all scools Ubuntu running :-) [04:23] Do you want me to tell this people arent in the position for solving problems like mine ? [04:23] xperia2: no. I want to not have to manage them. [04:24] and why do you think all questions need to be answered by the support [04:24] who said they do? [04:24] lamont: let's say that I recently saw a well-known conference website go off-line because of broken glue ;) [04:24] JanC: when glue is broken, it is the problem. when it's not, it's a non-issue [04:25] of course [04:25] since most people are hosting DNS $ELSEWHERE, glue seldom factors in. esp when we add in "make sure that your secondary is not one of the hosts in your domain" [04:26] lamont: I don't want to feed the owners of domains that have primary & secondary pointing to the same host within the domain ;) [04:26] having said that, it's _FASTER_ if you have all your nameservers inzone. [04:27] sometimes they're tricky though... they have 2 IPs that are on the same host. "much better" :( [04:27] and then they move and forget to change the glue too [04:27] lamont: didnt you said that the people arent proffesional enoght to answer all questions if i am not wrong [04:27] The normal answerable question will be answered other question that are specific can be put into a que. [04:27] I dont accepr just your thinking that such a service is to expensive and the people are to stupid to answer questions that are posted [04:27] day for day in the ubuntu forums. [04:27] xperia2: no. I said that if they aren't, then my reputation suffers, if I'm offering it as a for-fee service [04:28] lamont: depends if they live to tell the tale [04:28] This is why all my advice includes a halon discharge test [04:28] twb: lol [04:28] I think the issue is that it's often difficult to say how much time is needed to solve a problem [04:30] Regarding time estimates, I have had success with a 3-tuple approach: a "best case" estimate, a "worst case" estimate, and an "expected case" estimate. [04:30] The distance between them indicates to manglement in a measurable way the amount of risk/uncertainty [04:30] sometimes it takes 3 hours to solve a problem with somebody on IRC, but i doubt they would want to pay an industry-conform rate for that :P [04:31] JanC: well, IRC is for community, not industry [04:31] and sometimes after those 3 hours I have to tell them I don't know the solution (or there isn't one); will they still pay me? ;) [04:32] twb: what i answer for free, I can also answer for payment ;) [04:33] and i doubt an SMB would be happy with "sorry there is no solution" and then pay ;) [04:33] * lamont wanders off with family, fun though this has been. [04:33] I've also found on small jobs the no pay rate is pretty high, so in the unusual event I do them now, it's always pay in advance. [04:33] JanC: they don't have to be happy [04:33] lamont: have fun [04:33] JanC: they just have to pay [04:33] JanC: That also solves the will they pay if you don't fix it problem. [04:34] ScottK: I think my employers do it by an up-front credit check in advance, and forward defaulters to a collection agency [04:34] For small jobs it's not worth the trouble. [04:34] Yeah [04:34] well, if they have to pay for no solution, they won't ask next time... [04:35] I'm a one person company, so I don't have an office staff to deal with such things. [04:35] ScottK: that's why I work in a company [04:35] The company exists to insulate me from the "normal" people [04:35] lol [04:35] * ScottK is getting close to a decade of not having worked for a company [04:36] Technically I do work for a company, but I'm it's owner and only employee [04:36] It means my income tax form auto-completes and stuff [04:36] Heh [04:36] that would be the same if you'rethe only employee of your own company ;) [04:37] lamont: sorry but you want just offer a high class expensive service for people that dont want that. most people have specific question like how can i configure conky or install it. what for repos do i need to activate and such a thing. for exactly such question people need a service package and dont tell me that the actuall solution of ubuntu is really good ! [04:38] how many question over and over are posted in ubuntu ? [04:38] xperia2: IIRC Canonical has end-user support now? [04:39] JanC: they have a call center in .ca somewhere [04:39] JanC: That's where this started. [04:39] xperia2 wants a less expensive support option for home server users [04:39] how many times did you read in the forums "Can you please read the wikipage .." and such on [04:39] xperia2: I don't want to offer any for-fee support service to anyone for any rate. [04:40] I'm convinced that home users aren't worth supporting [04:40] * ScottK neither [04:40] * ScottK has had several business ideas in the last few years that he didn't pursue due to needing to provide user support. [04:40] They are too poor, too argumentative and too stupid. [04:40] Lamont: i know ! i am talknig more generally [04:40] then don't tell me that I want to. [04:41] I especially like working for government [04:41] Canonical and others are happy to provide for-fee services that they see a business model for. If you see a business model in other things, well, go for it. [04:41] Because $boss spends a month getting the contract, and then I get a nice big project to work on. [04:41] Lamont: :-) okay [04:43] ScottK: How do you land your support jobs if you are the only employee?? like how do you market .. etc.? [04:44] Most of the projects I work on tend to be large, long projects and word of mouth does pretty well for me for marketing. [04:44] actually, I love to work on end-user problems, I just doubt I can live on that... ;) [04:45] lamont: Oooh, MPE. There's a blast from the past. [04:51] wanted to tell only that i have found only that dell offers end-user support fo ubuntu for low price but didnt find that ubuntu is offering end-user support [04:54] I didn't know Dell even shipped Ubuntu [04:54] Desktops, not server. [04:55] Like I said [04:58] are there any desktop guis for managing server services like apache, mysql, firewall, etc... all in one like windows or open suse? [05:00] ksoviero: what exactly do you want to make with this guis ? starting stoping apache only ? [05:01] for what do you need it ? [05:01] the ubuntu server boot the way exactly how i need it and should be ! why does it not fit for you ? [05:02] more than that, managing virtual server, enabling and disabling modules, etc... [05:03] with mysql, user managment, create and delete databases, create and edit tables [05:03] well for mysql you can use very good phpmyadmin [05:03] it is very easy for using [05:04] web based wont work, for reasons I wont go into , it needs to be desktop based [05:05] right now, webmin fits my needs perfectly, but, not I need something just like webmin, but desktop based [05:07] ksoviero: dont have experience with such desktop based guis for a server. best will be probably to ask in the mysql irc chanell [05:07] ok, thank you [05:08] i am wondering if mysql even could have such a support [05:09] i am running my server with no screen and connect over ssh to the server and can do all thing on the laptop over the internet conncetion. [05:15] okay people. i need to do my works. want to thanks again for your help. see you again. bye [06:38] in the openldap-server ubuntu server guide there are a few lines that are confusing me. Item 4 says to edit the following file /tmp/ldif_output/cn=config/cn=schema/cn={8}misc.ldif and change a couple of lines but they don't say what to change them to.. could someone give me a hand. im trying to set up openldap and have been stuck for a while [06:51] New bug: #481752 in bind9 (main) "package bind9 1:9.6.1.dfsg.P1-3 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/481752 [07:03] anyone here use ebox? for some reason the ebox-mail package doesn't exist [08:19] memort stick wont show up in /dev as sdb [08:20] when I try to mount it says device does not exist [08:54] can i install sever edition from hard disk? [09:19] gehengwang: I dont think you can, you need a cd [09:21] ESEDU: are you sure? [09:21] gehengwang: why, cant u just get a cd [09:22] ESEDU: where can i get it? [09:22] U can order it for free from canonical home page, it takes for about 2 weeks to arrive [09:23] or download the image for it from ubuntu pages and burn it on a cd [09:23] ESEDU: and my CD-ROM doesn't always work better! [09:24] gehengwang: / [09:24] : /* [09:24] ESEDU: ok, thanks! [09:24] ESEDU: / [09:24] ? [09:24] i was supposed to draw a sad smiley : ( [09:25] oh [09:25] an i also have a question! [09:26] ESEDU: why desktop edition can install from hard disk? [09:28] gehengwang: beats me. I to tried to install ubuntu server from hard drive, but some guy i recall saying that u cant, that u need a cd [09:30] ESEDU: ok .but thanks [09:30] np [09:30] u might want to ask some else still, for im no hack :P [09:31] ESEDU: ok === sleeping`dragon is now known as error404notfound [10:05] gehengwang: you can always do a network install [10:06] how can i start a process with a lower priority? [10:09] [related to mail management] does anyone know how to implement a functionality so when a user sends a mail with a big attachment it is parsed by the MTA, generated a link to a website and mail to the receiver with that link? [10:09] acalvo: can you tell me the details [10:09] ok got it..nice command.. [10:11] alex88: niceness maybe? [10:12] gehengwang: AFAIK, you can boot a system using a USB stick, or even a boot CD, and download all packages from internet [10:13] acalvo: it's "nice -n priority Command"..thanks anyway === Blusion is now known as Blusion`AFk === Blusion`AFk is now known as Blusion === georg is now known as kwork [11:06] <\sh> moins [11:08] I'm trying to create a script (it's bash but it could be perl) to connect thru ssh to a machine an execute a command. However, this script has to be called by one user which does not have home directory (is the nagios user), so it does not have any ssh related file. I've point the ssh command to take a public rsa, but if I try to connect it asks for the authenticity of the host, and since is running in batch mode I cannot acc [11:08] ept nor reject it. Any solution? [11:14] create a nagios user on the remote machine [11:14] why? the script tries to ssh using the root user [11:14] omg [11:14] why on earth would you do that??? [11:14] thus I can restart services on the remote machine [11:15] first, never let anyone in as the root user on ssh in the first place [11:15] man sudo [11:15] sudo does not work [11:15] I've tried that first [11:15] erm. it doesn't? [11:15] I mean [11:15] it works [11:15] what makes it _not_ work? [11:15] but if you run it as nagios it won't start the services because it does not launch the process as the user it should, and then it can't read the configuration files [11:16] opening up for root login over ssh is bad. automating root login over ssh is insane [11:16] sudo /etc/init.d/whateverprocess restart [11:16] yeah some other user with sudo with nopasswd [11:16] I didn't find a better solution [11:17] first try was to modify sudoers file and add the nagios user to that service [11:17] you can do everything thru sudo [11:17] however [11:17] I want to run remote cmds [11:17] so I've to set up in the remote machines [11:17] acalvo: make a script for what you want to do, sudo-run that script [11:18] nice point, didn't think of that! [11:18] RoyK: even if I'm running a ssh script? [11:18] then in sudoers, allow only that script [11:18] even if you're doing a bloody ls command [11:18] RoyK, i think you need to enable these commands to that user aswell ? [11:18] RoyK, whats run inside script [11:18] just hte script [11:19] s/hte/the/ [11:19] let's see [11:19] RoyK, strange, it didnot work for me like that [11:19] RoyK, should retest it [11:19] well, works for me (tm) [11:19] RoyK, thou it was openbsd maybe it was because of that [11:20] * RoyK hands kwork a -gh [11:20] no. sudo works the same across platforms [11:21] -gh whats that [11:21] New bug: #481847 in postfix (main) "package postfix 2.6.5-3 failed to install/upgrade: subprocess installed post-installation script returned error exit status 75" [Undecided,New] https://launchpad.net/bugs/481847 [11:21] kwork: just picking on your spelling [11:21] RoyK, sorry im not native speaker [11:21] me neither [11:21] http://www.goenglish.com/PeopleWhoLiveInGlassHousesShouldNotThrowStones.asp [11:22] RoyK, and irc has ruined my spelling to lazy level writing [11:22] but though is spelt though although some americans tend to spell though as tho and through as thru and so on, but fuck them :) [11:22] in america, spelt is not a word [11:22] at least not a verb [11:23] spelt, past form of 'spell' [11:23] http://www.spelt.com/ [11:23] he grain is naturally high in fiber, and contain significantly more protein than wheat [11:23] I don't care if Americans can't type as long as some English people actually can [11:23] and that's how the internet collapsed [11:24] RoyK: spelled, past tense of spell, in America [11:24] sorry, i just noticed this was not #defocus [11:24] Boohbah: americans speak and write fucked up English. I don't use US english [11:38] is not working [11:38] the script gets executed [11:39] but it does not restart the service [11:39] what happens if you try to run the script manually using sudo, from the nagios user? [11:39] this is the script [11:39] http://paste.ubuntu.com/317711/ [11:40] it works if I run it as the root user [11:41] is it possible to stop a package trying to configure itself when you apt-get install it? [11:42] erm [11:42] acalvo: why do you ssh further in the script? [11:43] to restart a service in a remote machine [11:43] what I do is I have a nagios user all over the boxes, ssh from nagios to nagios and run a script on the remote host [11:43] ssh nagios@somehost /run/this/script/on/the/host [11:43] on that host, I configure sudo [11:43] not on the nagios host [11:44] ssh nagios@somehost sudo /etc/init.d/bloodyservice restart [11:44] for instance [11:45] so I've to create a pair of public/private keys to ssh without asking for a password, right? [11:46] ssh-keygen [11:46] create the pair [11:46] copy .ssh/id_rsa.pub to the remote host's .ssh/authorized_keys [11:46] but how? the nagios user does not have a home folder nor a passwd, so I cannot su to that user [11:46] then create one! [11:48] since it is a service user, I don't think is a good idea modifying it [11:48] I don't want to break any further update [11:48] it's just a bloody user [11:48] an entry in /etc/passwd [11:48] etc [11:48] RoyK: it's a ldap based system [11:48] it's not as easy as /etc/passwd [11:48] but this isn't really an ubuntu-server question, /j #nagios [12:07] Anyone around with a knowledge of postfix? [12:07] I desperately need hints on how to get spamassassin to work with it [12:08] mok0: What's the problem? [12:08] ScottK: Spamassassin doesn't mark the mails, as if it doesn't see them [12:08] mok0: Generally the recommendation is to integrate amavisd-new with postfix and have amavisd-new talk to spamassassin [12:09] mok0: amavisd-new will take care of that. [12:09] ScottK: I thought amavisd was a virus scanner [12:09] So, the package to install is amavisd-new ? [12:10] mok0: No. It has some checks of it's own, but is primarily an integration point for other things like spamassassin and (if you want) clamav. [12:10] Yes [12:10] mok0: Which Ubuntu release is your server on? [12:10] ScottK: karmic :-) [12:11] ScottK: I'm moving the old mail server to a new setup, and I want to use Ubuntu's mailstack [12:11] mok0: https://help.ubuntu.com/9.10/serverguide/C/mail-filtering.html [12:11] ScottK: thanks [12:11] You're welcome. [12:12] ScottK: cool! The other guides I've seen on help.ubuntu.com are outdated [12:12] mok0: /j #spamassassin [12:13] or #postfix [12:13] RoyK: Thanks, but I actually know sa quite well. It's postfix that's new to me [12:13] heh ok [12:31] ScottK: there's nothing on that page telling you to integrate amavisd with spamassissin. Does that work out of the box? [12:33] mok0: It's the changes in /etc/amavis/conf.d/15-content_filter_mode that do it. [12:33] ScottK: right... I can see that it works now! [12:34] OK. Great. [12:36] ScottK: I think it would be good to have that whole mail chain set up in the dovecot-postfix package... [12:37] ScottK: ... with the guidelines telling you how to _disable_ the parts you don't want [12:37] mok0: I agree. I proposed the integration include spam/AV stuff, but postfix-dovecot was a first step. [12:37] ScottK: and a really helpful first step! [12:38] ScottK: it seems there weren't much site-specific stuff in the changes I just made to the conf files [12:38] No. I'll add a spec for UDS on it. I can at least ask. [12:38] ScottK: Great! [12:44] ScottK: that was fast! :-) [12:44] Doesn't take long to make a stub of a spec. [12:46] I'm quite interested in setting up spam/AV with postifx-dovecot [12:46] I'm stuck at having SASL installed [12:46] configured, I meant [12:46] acalvo: I didn't do that step [12:46] i am running ubuntu server 8.10 and i have a backup script running as a cron job. the problem is that it does not not backup all the files as a cron, but when i run it manually it it backsup everything [12:47] New bug: #482086 in ntp (main) "System time is very slow and differs with real for about 1 minute per hour" [Undecided,New] https://launchpad.net/bugs/482086 [12:49] ccole1: see man 5 crontab Sometimes you have to escape certain characters from the command cron runs. At least % signs are interpreted by cron [12:50] Especially date command format strings could cause this to happen [12:56] @hvn my crontab entry is this 05 12 * * * /backup/backup.sh. it does not have any of the things mentioned that might cause problems [12:59] hvn: my crontab entry is this 05 12 * * * /backup/backup.sh. it does not have any of the things mentioned that might cause problems [13:02] What is the location of the crontab? /etc/cron.d/ /etc/crontab or does it belong to a user directly? [13:03] looks like your example does not have the user specified, so it probably is not, or should not be, under /etc/ [13:03] I access it by using the command sudo crontab -e so i am making the assumption that it is the root user crontab [13:05] ok, I think that is the case. Does /var/log/syslog or other log files show anything related to the command? [13:05] just that it runs [13:05] no errors [13:06] it backs up my files only not the other users [13:06] I have to admit that I have not used other than /etc/cron.* locations for running cronjobs [13:07] Hmm, sounds like it does not run as root [13:07] let me check [13:08] when i check the details of the cron under webmin. it states that it runs as root [13:08] is there any other way to confirm this [13:11] I just tried sudo crontab -e on a 8.04. The crontab ends up as /var/spool/cron/crontabs/root and is owned by root:crontab [13:11] From that it looks like it should run as root [13:11] New bug: #480618 in dhcp3 (main) "package dhcp3-server 3.1.1-1ubuntu2.1 failed to install/upgrade: EOF da stdin al prompt per il file di configurazione" [Low,Incomplete] https://launchpad.net/bugs/480618 [13:13] ccole1: are the user dirs on an NFS mounted drive? [13:14] the user directories are on an ext3 filesystem [13:14] ccole1: ... local on the machine running the cron job? [13:15] ccole1: because your server may not have root priveledges on a disk drive from a remote machine [13:15] the ubuntu machine is my file server using samba i connect each win xp machine to the server [13:16] so the cron job runs local on the server, and all files are store locally on the server [13:16] ccole1: alright... just a thought [13:17] If you do sudo ls -l /var/spool/cron/crontabs/ do you see the crontab owned by root:crontab? [13:18] yeah [13:19] Nov 5 12:05:01 lostcity /USR/SBIN/CRON[7818]: (root) CMD (/backup/backup.sh) [13:19] the above is a line from my syslog [13:20] ccole1: ought to work [13:20] ccole1: get your script to output some messages about what it's doing [13:22] i have it outputting all messages it does not complain about anything it just does my directory and ignores everyone else [13:22] could it be the directory permissions [13:22] ok. Then I would check the script and check what environment variables, including contents of PATH, it thinks are available. The environment available to cron may not contain everything the script needs espcially when it is passed from cron to the script it launces. [13:22] ccole1: your script is not executed under the same environment as root. Specifically, PATH might be different [13:23] hehe [13:23] hvn, beat me to it [13:23] just slow typing :) [13:23] hvn, ... or thinking :-) [13:24] i have set a directory mode of 700 for everyone, could that be the problem [13:24] ttx: ping [13:24] zul: pongpong [13:25] if it runs as root, the permissions should not matter [13:26] ttx: we have the same version of net-snmp for the last 3 releases of ubuntu i was thinking about looking at the version in debian/testing for lucid what do you think? [13:26] i have put my script up on pastebin the url is http://paste.ubuntu.com/317783/ [13:28] output from the err log for this script is at http://paste.ubuntu.com/317784/ [13:29] it runs perfectly when i run it manually [13:29] ccole1: uhm... why don't you just use rsnapshot... it's much better than your script :-) [13:29] ccole1: much less overhead and can save N days of backup [13:30] ccole1: takes about 5 minutes to set up, can backup to a local or remote system [13:33] it feels like a cop out, like i could not get a simple script to work so instead i ran away [13:34] ccole1: once you see what rsnapshot can do for you, you'll get over your hurt pride :-) [13:34] zul: yes, good idea [13:35] not a cop out.. nothing wrong with using code someone else wrote that already does exactly what you need. After all, you're already using open source... you couldnt' build your own kernel so you ran away. :) [13:35] The first line of script is not e.g., #!/bin/sh [13:36] hvn: is that important? I think cron spawns the shell... but good catch [13:36] I am not sure how the executable to run it is chosen [13:36] and if the pastebin errlog is from cron, it runs [13:37] let me try it now === _jakob_ is now known as Synthesis [13:38] what's that "2>>" doing? [13:39] I am trying to install Ubuntu 9.10 x64 on HP bl460c, everything goes fine during the install, but when i reboot, Grub cannot find the kernels or anything under /boot... Any suggestions ??? [13:39] kirkland: *cough* there is seven new qemu-kvm bugs sitting there for you [13:39] it directs stderr to the file instead of the screen [13:39] I can never remember these weird redirects [13:40] Looks like C++ code :-P [13:40] redirecting stderr. that is why there is "Removing leading '/'" and not file list in the log [13:40] I'm trying to manage a service using another user. I've set up the sudoers file to let the user manage the init.d script, so it can fire it, but I'm getting errors with file permissions in the config files of the service. [13:42] the file list resulting from v-option is going to stdout and ending up in email sent to root (if there is enough mailer installed) [13:42] I still say: go for rsnapshot: http://rsnapshot.org/ [13:42] apt-gettable [13:44] If you're clever, you can set it up so users have access to their own backups [13:46] New bug: #481203 in openssh (main) "Hardy 8.04.3 openssh does not forward X11 always cannot open display" [Low,Incomplete] https://launchpad.net/bugs/481203 [13:47] i added the #!/bin/sh. it run as before no dice [13:49] I see that you have commented out line 20. If you enable it, does the file list from tar look as it should? [13:51] yeah it just list all the files that it is copying. that is why i took it out i wanted to focus on errors alone [13:52] And that file list shows that not all home directories are traversed? [13:53] yes, it just traverses mine, and then continues on. it does not show any sort of error, which is vexing [13:56] There's a strange thing: I can mount my raid5 (which I set up in a previous install using mdadm) using mount /dev/md0 /mnt/raid, but when I put /dev/md0 /mnt/raid xfs 0 0 in fstab, I get a bad-superblock error! [13:57] Can it be a mdadm config thing? because I know I had to put some information there about the raid, but don't find the howto anymore.. [14:01] ccole1: I sounds like your home directory is the only one in /home If tar can not access files, it logs an error. Since there are no errors, it can access everything. Are you sure your /home is where the others have their home directories. Sorry if this obvious :) [14:01] It's the straw grasphing momement :) [14:01] Just a quick check. Disk space is not a problem? [14:02] ls -all /home reports everone in home [14:03] webmin reports about 8% disk space used [14:07] Interesting problem. Can't really think why it is behaving like that. [14:09] How can I have the tmpfs use less RAM? I have 80gb of ram and tmpfs used 5 partitions of 12gb each, leaving only 24gb of ram! [14:09] 80gb of ram ? [14:09] And they're all almost empty, while my processes get killed because they don't have enough ram... :( [14:10] me either and it is drving me crazy [14:10] kwork: yes, it's a nice server! I got it as a gift from ETH biochemistry.. :) [14:10] thats a lot of ram [14:10] nyk2005, wiht kind of hardware is that [14:10] what [14:11] but no need to make tmpfs out of all of it, the server also has 80gb of SD hdd which is almost as fast.. [14:11] http://biodata2.ethz.ch/phpsysinfo [14:11] I'm running hardy on a Pentium III 450 (and corresponding old motherboard) with USB1. Would I be able to install a PCI USB card and get USB2 speeds? [14:11] ccole1: the tar command you are using. There is no chance that it is something else than /bin/tar? [14:12] I want to connect and external HD and make my own "NAS". [14:12] There are no scripts or anything that run instead of /bin/tar [14:12] let me check [14:15] used sudo find / -name tar, it only turned up /bin/tar [14:15] nyk2005, sysinfo shows only 24gb [14:16] nyk2005, nice [14:17] kwork: because the rest is used by tmpfs? That's why I'm asking about reducing that wastefull config! [14:18] hmmmz indeed [14:18] wicked config :D [14:18] 5*12+24 =about= 80 [14:19] and those tmpfs are a waste of ram... the admin who gave me the machine already wondered what this "linux thing" is doing... :) [14:19] nyk2005: it swapped out, doesn't take any ram [14:20] And searching for ubuntu reduce size tmpfs in google only brings up lots of peples question about not understand what tmpfs is.. [14:20] mok0: really? [14:20] yes [14:20] mok0: but one of my R processes was killed last night! [14:20] nyk2005: why? [14:20] nyk2005: lack of scratch space? [14:21] mok0: let me check the log.. [14:21] nyk2005: http://en.wikipedia.org/wiki/Tmpfs see the section on Linux [14:22] nyk2005: it's pretty much standard in modern OSes [14:22] mok0: the log end with "Killed". This only happens if the kernel has to kill something because it uses all RAM or some other trigger... [14:23] nyk2005, could you paste lshw total [14:23] during mapping an exonarray, which takes some gigs of ram.. [14:23] of memory [14:23] lshw -C [14:23] lshw -C memory [14:23] arg tunnels.. [14:24] ok [14:24] just the last? [14:24] theres total somewhere there [14:25] http://biodata2.ethz.ch/lshw.txt [14:25] !pastebin > nyk2005 [14:25] nyk2005, please see my private message [14:25] mok0, ? [14:25] nyk2005, it seems its 24gb [14:25] yeah? [14:25] !pastebin [14:25] pastebin is a service to post multiple-lined texts so you don't flood the channel. Ubuntu pastebin is at http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic [14:26] mok0, why use pastebin over nano /var/www/somefile [14:26] ah ye [14:26] ah yes [14:26] ccole1: Got me with the backup thing :( [14:26] yeah I have a webserver just for that file.. :) [14:27] kwork: pastebinit < file [14:27] dunno i dont like to use pastebin at all [14:27] Can't say what it could be apart from logging in and checking it myself. And that is not an option :) [14:27] so it has 24gb of ram and the same amount of ram used by tmpfs sharedly.. strange, I saw it going to 80gb in bios.. [14:27] why would i want my configs to be floating around in some random server for who knows how long ? [14:27] nyk2005, what kind of connection do you have on this server ? [14:27] kwork: I'll delete the file.. [14:28] nyk2005, kk [14:29] well thanks for all your help! I run the process again see if it gets killed again. after all last time I tried the install was broken. [14:29] thanks [14:30] nyk2005: if your app uses /tmp extensively, and at the same time you do not have much RAM, your system could run out. [14:30] orudie_: why? I have no idea... good one I guess, it's located in the newest server-room of the ETH in the new HIT building. the room looked like science fiction to me... servers almost frozen.. [14:30] nyk2005: you should configure your servers with plenty of swap space [14:31] mok0: why? /tmp is on root... I though also it should at least be on tmpfs.. [14:31] nyk2005: /tmp is a filesystem mounted on / [14:32] nyk2005: It takes no disk space away from / [14:32] and I guess all will be ok, mainly I need a lot of ram and not so much /tmp.. [14:33] mok0: really? why doesn't this show up in fdisk or mtab then? [14:33] nyk2005: not sure about R... does it have proper garbage collection? [14:33] nyk2005: eerr .... I am talking about the default Ubuntu setup. Perhaps you changed it? [14:33] mok0: well depends probably on the package.. [14:34] I guess [14:34] mok0: nope, just installed default ubuntu-server [14:34] Hm, you [14:34] are right... [14:35] nyk2005, i cant ping your domain biodata2.ethz.ch [14:35] nyk2005, or your ip [14:35] yeah if it fails again in R then it's probably that I'm using the package in an inefficient way. thanks for all the help, just wanted to be sure that the hardware is all ok.. [14:35] orudie_: it's of course firewalled [14:36] nyk2005: why? [14:36] nyk2005: what harm can ICMP echo packets do? [14:36] yeah you should be able to ping your server [14:36] just like ping google.com [14:36] Boohbah: what good can it do me if someone pings my server? [14:36] soren: ping [14:37] no... [14:37] ah well... I wouldn't care honestly [14:37] it's the admins that control the firewall and I'm happy to have http/https/ssh open [14:38] it would let them know that your server is up [14:38] the local admin knows.. [14:39] where is the server located ? [14:39] why ping? [14:39] orudie_: zurich [14:39] oh [14:39] why? [14:40] you're there too? [14:40] :) [14:40] no [14:40] i'm in New York [14:40] my ubuntu server is in Atlanta, Georgia [14:40] oh! [14:40] I'm also not in zurich anymore but in the train to bern... : [14:40] now that it works finally [14:41] bern ? [14:41] lets give it something to work on! [14:41] yes, capital of switzerland [14:41] oh [14:41] are you in the moving train right now ? [14:42] yes [14:42] in a tunnel now.. :) [14:43] what is the primary purpose of the server, why does it need so much resources ? [14:45] orudie_: analysis of mouse exon arrays and comparison to other data. running gigantic mysql and psql dbs [14:47] orudie_: and well... anything I come up with. :) [14:47] ttx, around? [14:48] jcastro: yes [14:48] I've given you scheduling powers in the summit system [14:48] I've been told you will be scheduling the server bits? [14:48] jcastro: who told you that ? [14:49] jcastro: mdz asked to schdule them himself. [14:49] jono did, and for me to help you? [14:49] oh [14:49] ok then [14:49] well, if worse comes to worse, you have the right rights to adjust thing [14:49] jcastro: I might have missed the memo though :) [14:49] heh [14:49] * zul bows down to our new overlord [14:49] ok so just in case he gets hit by a bus [14:49] jcastro: I had some rights already, not sure that was enough though [14:49] you just login to summit.ubuntu.com [14:49] and you can drag sessions around [14:50] jcastro: inherited from when I handled the invitations [14:50] jcastro: ok [14:50] ttx: I'd appreciate getting https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-more-mail-integration scheduled. [14:50] ttx: and sever-lucid-papercuts ;) [14:51] ScottK: like I said, so far I've been asked to stay away from the scheduling, so I'll wait for the memo [14:51] Ah. I misread. [14:51] jcastro: ^^^ would you please schedule https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-more-mail-integration then. [14:51] ScottK: but I'll gladly help if I get the authorization to do so :) [14:51] Understand. [14:52] yeah, ttx or mdz need to schedule it [14:52] I don't touch tracks without the track lead telling me so [14:52] jcastro: same here :P [14:52] OK. Since ttx isn't scheduling, then we wait for mdz. [14:53] ttx, I was told you were backup man. ;) [14:53] but ok [14:54] ScottK, jono has authorized me to schedule your motu one at least [14:54] jcastro: we got different messages. [14:54] jcastro: Thanks. [14:54] nyk2005, pm [15:07] would anyone happen to know why when using sftp, a user can create a directory and chmod the directory but a chgrp will fail with unable to [15:07] setstat permission denied? [15:13] ttx, jcastro, I've asked ttx to take responsibility for the second "roomful" of sessions [15:14] oh ok, so if there's a sever session scheduled but you need a second one in the slot then just handle it? [15:14] easy enough [15:14] mdz: ah, ok. [15:15] mdz: in a recent email ? [15:20] kirkland, mathiaz: about eucalyptus -proposed validation: I'd need to do a complete reinstall to validate 7 -> 7.3 upgrade path and it looks like I won't have time to do so. Could you both have a look and validate the fixes if possible ? [15:20] kirkland: you'll bring the cloud-in-a-bag in Dallas, right ? [15:20] ttx: i will bring the cloud with me [15:20] ttx: i'm on holiday today [15:21] kirkland: ah sorry :) [15:22] ttx: i will bring the cloud with me <-- sounds like David Lynch [15:23] I won't be able to go to UDS, but I hope that for the European folks who are going for the first time - please do try the barbeque. Texas is known for their beef brisket. Here in Alabama, it's usually pork. [15:23] * ttx looks at the schedule [15:23] kirkland: what machines did you use to make the "cloud in a bag" ? [15:24] aubre: laptops [15:24] kirkland: I know, I was wondering what models [15:25] aubre: dell vostro 1220's and thinkpad x61's [15:25] tyvm [15:26] kirkland: I am excited to be trying testdrive, it looks like a wonderful project [15:29] aubre: cool! glad you like [15:31] kirkland: I like the fact that when you go to look at the Vostro they make a point of saying that the CPU has VT support [15:32] aubre: virtualized netbook... sounds... not useful. [15:33] Boohbah: it is if you are showing someone UEC [15:33] ttx, in an email I hadn't sent yet ;-) [15:33] Boohbah: and you need a portable cloud [15:33] mdz: ah :) [15:33] Uranium Energy Corp? [15:33] Utah eLearning Connection? [15:33] Boohbah: Ubuntu Enterprise Cloud [15:34] https://help.ubuntu.com/community/UEC [15:38] ScottK: the mail session is not "accepted" for UDS yet, and I don't have the power to do so (yet) [15:38] jcastro, mdz: ^ [15:39] https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-more-mail-integration [15:42] ttx, you need to be put in the uds-organizers lp group, only a tech board member can put you in that group. [15:42] in order to accept sessions for uds [15:42] jcastro: ok :) [15:44] smoser: ping [15:44] here [15:51] smoser: you proposed the following sessions on the secondary track: === dendro-afk is now known as dendrobates [15:52] server-lucid-cloud-krd, server-lucid-cloud-ovf [15:52] do you agree to chair those at UDS ? [15:53] yeah. thats fine. === robbiew_ is now known as robbiew [15:57] I have a mysql-server running, but when I move it's data dir from /var/lib to /mnt/raid (on a mounted raid5) it refuses to run as it can't create any files. But normal users can write there. [15:59] im having trouble getting my bcm4312 card working with the new karmic server kernel [16:00] i've tried reinstalling the bcmwl-kernel-source modalias and jockey but dkms fails === mushroomtwo is now known as mushroomblue [16:00] is there anyway to tell my server that "myhost.com" is "192.xxx.xxx.xxx" if I connect through eth0, and it's "198.yyy.yyy.yyy" if I connect through eth1? [16:00] reinstall of bcmwl --> http://paste.ubuntu.com/317861/ [16:01] jockey log --> http://paste.ubuntu.com/317873/ [16:04] will be on and off if someone know how to manually compile and do what dkms does and can point me to a tutorial that would be nifty [16:08] I have a mysql-server running, but when I move it's data dir from /var/lib to /mnt/raid (on a mounted raid5) it refuses to run as it can't create any files. But the normal users can write there! What's wrong? [16:10] nyk2005: did you keep the owners and permissions? [16:11] Reepicheep: yes [16:11] and I assume you updated the "datadir" in my.cnf [16:13] yes [16:13] strange, after chown root:root /mnt/raid/mysql/mysql_upgrade_info it worked. [16:13] as if this had to belong to root.. [16:14] nyk2005: what command did you use to copy the data? [16:14] cp -r [16:14] then chown -R mysql:mysql [16:15] cp -r doesn't keep owner and permissions [16:15] you may of needed to use "cp -ra" instead [16:16] it doesn't work anymore now.. :( [16:16] mush have been coincidence [16:16] nyk2005: so it was working but now it isn't working? [16:17] Reepicheep: yes... very strange [16:17] you still have the data in /var/lib/mysql correct? [16:17] and I just deleted /var/lib/mysql [16:18] nyk2005: what is the error now? [16:21] Reepicheep: just "fail" [16:21] Reepicheep: nothing in /var/log/messages [16:22] Reepicheep: only a lot of "Nov 13 17:20:44 biodata2 kernel: [10484.346455] type=1503 audit(1258129244.995:475): operation="open" pid=18365 parent=18364 profile="/usr/sbin/mysqld" requested_mask="r::" denied_mask="r::" fsuid=0 ouid=0 name="/sys/devices/system/cpu/" [16:31] similar problem: http://ubuntuforums.org/showthread.php?t=928445 [16:32] yes its know [16:33] what? [16:34] ah: you also need to also adjust /etc/apparmor.d/usr.sbin.mysqld. [16:37] nyk2005: try starting it with mysqld_safe [16:37] finally it works! apparmor ... is annoying [16:38] the /sys/devices/system/cpu/ should be cosmetic [16:47] how do I expand my cloud with a second cluster controller/availability zone? I installed eucalyptus-cc and eucalyptus-sc on a host, then changed the /etc/eucalyptus/eucalyptus-cc.conf with a second availability zone name, and eucalyptus-ipaddr.conf with the second cluster controllers IP [16:47] however euca-describe-availability-zones verbose only shows the original avail. zone [17:08] What could cause rsync to suddenly hang? I've copied several directories, but suddenly both rsync and ssh hang. [17:09] Even ssh remote ls doesn [17:09] t work [17:12] ping? [17:31] ttx: hey [17:31] ttx: I'll do the upgrade SRU validation [17:31] ttx: ok, the server sessions are all marked now, should be colored [17:45] <_markh_> Hi everyone. Just starting to learn about clouds with ubuntu. In the docs, it states that the front-end must exist on the same class C net as the nodes. Is this true? If so, how can I set up a cloud that mixes private nodes and public space (say EC2)? Surely such a constraint would mean that I also couldn't easily have hosts from different server providers participating in a single cloud. [17:47] is there a solaris live upgrade like feature on ubuntu? where you detach the mirror and then upgrade the OS on the detached inactive sub mirror device and boot from that .. and if something is wrong then you boot back to current boot env ? [17:52] axisys: The closest thing available that I'm aware of is ksplice [17:53] kirkland: when you get a chance, rcs_cost branch at lp:~jbernard/+junk/byobu [18:25] _markh_: by the same class "C" network it refers to that fact that the Cloud Controller and the Node each need to be able to communicate directly with each other on the same network [18:26] at least that's how I interpreted it. You can have multiple interfaces on the controller and the nodes though [18:27] so they can communicate on a private network.. but yet the virtual machines get access to a different public network [18:27] or how every you configure it. [18:27] <_markh_> Reepicheep: Why the same network, isn't it sufficient for them just to be able to talk to each other through any route? Or does the controller broadcast data? [18:28] I'm not sure on that .. but I think it is because the nodes auto discover the controller [18:28] so yeah.. probably a broadcast [18:29] see: http://open.eucalyptus.com/wiki/EucalyptusNetworking_v1.6 for more info on the networking [18:29] you may need to use one of the "Managed" network modes [18:31] <_markh_> I'll check that link in a moment - thx. In my setup I am setting up a test cloud with two hardware machine. Host A has IP addresses 80.84.51.34 - 80.84.51.38 and 192.168.0.1, Host B has addresses 80.84.54.98-80.84.54.102 and 192.168.0.2. I'm at a loss to know how to set (address wise) up A as the controller and B as the node [18:39] Can someone help me with cups? [18:40] <_markh_> Reepicheep: Got to dash (weekend here:) Thanks. [18:40] I am having issue connecting to it using the web management interface on a remote machine. === robbiew is now known as robbiew_ [18:56] I upgraded from jaunty 64bit server to karmic 64bit server.. and now I do get the login prompt in console.. this page https://help.ubuntu.com/community/SerialConsoleHowto says to edit /etc/default/grub .. but I dont see that file.. [18:57] I meant, I do not get the login prompt in console [18:57] do I create the /etc/default/grub file ? [18:57] would be nice if upgrade took care of it [18:58] fresh install do not experience that issue [19:03] /etc/init/ttyS0.conf this did the trick on that file [19:03] should I be brave and create a default grub /etc/default/grub ? if that breaks grub i need to network boot and install grub again ? [19:18] Greetings and salutations. [19:18] I've just set up server 9.1 and am trying to configure rsyslogd. [19:18] There used to be /etc/default/rsyslog, but that seems to not exist [19:18] when running pgupgradecluster to move postgres from 8.2 to 8.3, the new cluster is being created with a default encoding of LATIN-1. I'd prefer this to be UTF-8. As I understand it, pgupgradecluster looks at some environment vars to guess the correct encoding. Does anyone know how I can change the default? [19:18] if I create that file will it supersede the option settings in /etc/init.d/rsyslog still? === dendrobates is now known as dendro-afk [19:35] genii: i think ksplice only prevents reboot [19:35] genii: after kernel upgrade [19:35] however that never worked for me .. i had it installed on my laptop and it requuired a reboot after that kernel upgrade [19:36] stgraber: ping [19:44] ttx: there is only one session about etckeeper? [19:44] mathiaz: yes. [19:44] ttx: server-lucid-puppet-etckeeper-integration? [19:45] yes [19:55] ttx: pong [19:56] stgraber: about the containers session [19:56] stgraber: are you ok to chair that one ? [20:04] anyone having experience with crechendo product ? [20:04] (crescendo) [20:05] when i install mount points on different encrypted disks, and enter the passphrase for the first disk at boot, other messages intterrupt me before i can enter subsequent passphrases, is there a way around this? [20:06] stgraber: ? [20:06] is it possible to switch those subsequnt disks to keyphrase mounting? [20:06] sorry keyfile [20:07] stgraber: please email me the answer :) === dendro-afk is now known as dendrobates [20:17] can one run an ssh server from the installer disk (for remote installs) ? [20:19] StrangeCharm_: Yes [20:19] you can select packages to install somewhere [20:19] Jeeves_, how? [20:19] use 'Back' one time and you get this menu somehow [20:19] (Assuming your in the server-installer) [20:19] openssh is one of the packages that can be installed [20:20] Don't have an installer to find out how exactly, but afaik, you can use ssh in the installer [20:20] off to TopGear now :) [20:21] Jeeves_, i don't want to ssh from the installer; i want to be able to ssh into the installer [20:21] StrangeCharm_: Ah, no clue than. [20:21] ok, I'm confused... where's the startup options for rsyslogd stored? [20:22] /etc/rsyslog.conf /etc/rsyslog.d/* [20:22] Ow [20:22] startup options [20:22] Actually that's per application over-rides. [20:22] The main config file is /etc/rsyslog.conf [20:23] startup is in /etc/init/rsyslog [20:23] which you shouldn't edit [20:23] once upon a time there used to be a /etc/default/rsyslog file with RSYSLOGD_OPTIONS="..." which overrode the ones in /etc/init.d/rsyslog [20:23] But I guess they missed a step in implenting upstart and keeping all the options :) [20:24] Fenix|work: They dropped /etc/init.d/ in Karmic [20:24] moved everything to /etc/init/ [20:24] Which are upstart files [20:24] Which you cannot configure, regression, if you'd ask me :) [20:24] so I can't edit the /etc/init/* files then [20:24] so how do I override rsyslogd to allow remote connections? :) [20:25] You can, but you shouldn't === mushroomb1ue is now known as mushroomblue [20:25] So I'd ask the package maintainer :) [20:26] Maintainer: Ubuntu Developers [20:27] Fenix|work: yuck. File a bug, /etc/default/rsyslog is still there, and should be honored. [20:27] sbeattie: It's not there. But you mean it should? [20:28] Hunh, it is here, but that may be due to me ugprading through the karmic cycle. [20:29] fresh install of karmic, no /etc/default/rsyslog [20:29] looking in /etc/init/rsyslog.conf ... exec rsyslogd -c4 ... hardcoded right in the file [20:29] * ScottK doesn't have one either (on an upgraded system) [20:30] odd, I have it on two different karmic systems. Anyway, it's definitely a regression in configurability. [20:31] -c4 is a new option for rsyslogd ... -c4 selects the desired backward compatibility mode... so -c4 means no backwards compatibility as it's version 4 [20:32] Fenix|work: I ran into the same issue. I opened Bug 465657 [20:32] Launchpad bug 465657 in rsyslog "Cannot Start UDP or TCP Listening for Logging From Remote Clients" [Undecided,New] https://launchpad.net/bugs/465657 [20:32] let me add my two cents [20:34] Fenix|work: I found that the -r option is no longer used. However, following the new procedure and uncommenting the two options to enable remote logging in /etc/rsyslog.conf does not work either. [20:36] Grrr === mushroomtwo is now known as mushroomblue [20:41] jsalisbury, uncommenting those two lines for udp and adding ModLoad imtcp , $InputTCPServerRun 514 ... netstat -vatun shows that tcp and udp are both listening to 514 [20:42] Fenix|work: so it works for you? Hmm, failed for me. I'll have to go back and retry it. [20:42] it shows that it's listening, but maybe fw is blocking, I haven't tried logging yet [20:42] anyone with bonding instructions handy? [20:42] Hi all. I am trying once again to set up a PDC with Samba. It is my understanding that Windows machines need to give the root uname/pw to join the domain on the first login. It is there some way to set the root password in Ubuntu without booting through the recovery console or some other way to auth the Windows machines on the domain? [20:42] Fenix|work: I couldn't even get it to start listening [20:43] service rsyslog restart and it was good to go [20:43] Fenix|work: Hmm. must be me then ;-) [20:44] is there any fancy way to bond in 9.10 now? :) [20:48] Fenix|work: Nice, its working for me now. Just talking with you about it fixed it :-) I guess I'll close that bug now. Sorry, I don't know about bonding in Karmic - yet ;-) === jussi01 is now known as jussi01_ [20:51] we'll see how this works === robbiew_ is now known as robbiew [21:14] jsalisbury, https://help.ubuntu.com/community/UbuntuBonding -- works with karmic [21:15] Fenix|work: Thanks! === jussio1 is now known as jussi01 [21:32] Ubuntu Server 9.10: How do I set up "shared physical device"? tried the wiki on networking for virtualization and its not working [21:40] anyone know how I can get pgupgradecluster to use a different encoding (UTF-8) than the one it's guessing I want based off of my system locale(LATIN-1)? [22:01] Can anyone tell me, is LDAP necessary for Samba to function as a Windows PDC? I am so lost in the reeds right now. [22:01] majuk: ldap is not necessary for that [22:01] ahasenack, Thanks. [22:02] Now if I could only figure out how to make it work. >< [22:43] is it possible to link up the snakeoil cert to postfix ? [22:45] like [22:45] root@starbug:~# ln -s /etc/ssl/certs/ssl-cert-snakeoil.pem /etc/ssl/certs/smtpd.crt [22:45] root@starbug:~# ln -s /etc/ssl/private/ssl-cert-snakeoil.key /etc/ssl/private/smtpd.key [22:48] ok that works good === dendrobates is now known as dendro-afk === genii__ is now known as genii [23:57] I posted in #ubuntu questions about installing a GUI to experiment in Ubuntu Server 9.04, previously I tried Webmin and dont care for it. I have also been directed to ebox, or just use the desktop distribution. Can anyone offer a good supported solution to use a GUI to configure and experiment with to learn?