=== CreativeEmbassy is now known as Judy === Judy is now known as CreativeEmbassy [04:28] this is very stange problem i come across i have a utm /firewall as a gateway (another system ) ,i have configure web server ,ssh server and ftp server on that but i am only able to acess ftp server ,not ssh and web server but i change hardisk with oracle linux i am able to access all 3 thing (web ,ssh,ftp ) [04:33] running iptables or ufw [04:34] use nmap to scan the machine [04:35] check with that also [04:35] ufw uninstalled [04:36] this is very stange problem i come across i have a utm /firewall as a gateway (another system ) ,i have configure web server ,ssh server and ftp server on that but i am only able to acess ftp server ,not ssh and web server but i change hardisk with oracle linux i am able to access all 3 thing (web ,ssh,ftp ) [04:36] sorry [04:40] check iptables: sudo iptables -L [04:42] yes ,check that also [04:43] disabled [04:43] and i even change the ports also of web and ssh [04:44] try to put ssh on the ftp port and see if it works [04:44] ok [04:44] it still sounds like some firewall rules between the box you are trying to connect to and the client box [04:45] don't forget to shutdown ftp service before binding ssh to the ftp port [04:45] yes [04:49] also make sure the serivces are starting correctly and listening: sudo netstat --tcp --udp --listening --program --numeric-ports | grep sshd [04:49] no its not working [04:50] sounds like a service issue if ftp worked and ssh didn't on the same port [04:51] never face this kind of problem [04:51] and hight is if i change hardisk to oracle linux all things works fine [04:53] I would put ssh back to port 22 and run sudo netstat --tcp --udp --listening --program --numeric-ports | grep sshd [04:54] yes check [04:54] i even chage ssh port 5123 [04:56] I would use nmap to see if it is listening....I bet from the remote machine it doesn't find ssh running [04:56] and something is blocking it in between [04:56] www.utmxtm.com [04:56] can yo pls try [04:57] any result [04:57] its running now [04:58] I wonder if your ISP just blocks certain things. [04:58] rdp and vnc not configure [04:58] tryed ? [04:59] no i dont think [04:59] just change hard disk [04:59] it all work with oracle linux [05:00] do you have the the default route ? [05:00] correct [05:00] yes i am hopeing that or metric is wrong but dont know how to check that [05:01] sudo route [05:01] the default value will point at your router usually like mine is 192.168.1.1 [05:01] default route is my utm box [05:01] yes that is route [05:02] nmap is slow...it will be done in a second [05:02] thesheff17: nmap -F [05:02] thx [05:02] thx ? [05:03] !thx [05:03] You're welcome! But keep in mind I'm just a bot ;-) [05:03] lol [05:03] says ftp and https is open nothing else [05:03] thesheff17: putting SSH on a nonstandard port will affect QoS attempts by your ISP et al. [05:03] port 21 and 443 [05:04] If you are concerned with brute-force attacks an IPS would IMO be better than relying on obscurity [05:04] this works [05:04] https://www.utmxtm.com [05:05] The page is empty though [05:05] what is this user portal? is that what you are using? [05:05] hmm that is my utm portal [05:06] http://cyber.com.au/~twb/doc/iptab.ips [05:07] if https://www.utmxtm.com/ is not your linux box you have to do port forwarding..... [05:07] can you try now https://www.utmxtm.com [05:08] yes i did that [05:08] utm box has DNAT rules for all servicess [05:08] can you try now https://www.utmxtm.com [05:09] samba35: just get a second box and test it yourself [05:09] yea you have bad port forwarding or firewalls rules. [05:10] Probably just triangle routing [05:10] ok [05:11] all things are same just change hard disk to oracle linux all works ! [05:11] linux on oracle and ubuntu are almost identical [05:11] dhcp assing ip to linux [05:11] dhcp should never be used for a server [05:12] ic [05:12] i dont know that [05:12] thesheff17: I disagree. [05:12] thank i will try to put staic ip and chaeck yes we have that options open [05:12] Using fixed DHCP allocations to non-core servers allows them to be configured on one end, rather than both ends. [05:13] It *does* mean the LAN they're on must be ultimately trusted, of course. [05:15] well it sounds like a firewall iptables port forwarding mess....when I hit his portal I knew that wasn't his ubuntu box....him setting up fixed DHCP sounds like another layer of complexity :) [05:16] Granted [05:16] It is certainly useful to switch to manual configuration during testing. [06:53] does freenode allow dcc [06:55] Dunno, ask #freenode [06:55] irc network does not need to support it [06:55] onre: it's just CTCP [06:55] onre: right? [06:55] it works so that clients send CTCP-style messages to each other and initiate a tcp connection between each other [06:55] yes [06:56] via CTCP they communicate IP addresses and port numbers [08:28] Hi === himcesjf1 is now known as himcesjf [08:31] I am using Ubuntu orchestra and I would like to get a pointer to configuring rsyslog. [08:31] I am referring this URL http://wiki.rsyslog.com/index.php/Very_simple_config_--_starting_point_for_modifications [08:31] Will this be ok ? [08:31] I am using Ubuntu Oneiric and basically want to log remotely to the Ubuntu Orchestration server from a node or nodes. [08:32] Thanks in advance. [08:32] The rsyslog is not configured by default in orchestra. [08:37] Vivek: did you install 'ubuntu-orchestra-logging-server' ? [08:39] Vivek: orchestra doesn't do much of the configuration for you.. its mostly just a collection of packages... to configure the resulting servers you need something like juju or puppet. [08:39] Open-source or closed-source for SaaS? - http://programmers.stackexchange.com/q/130341 [08:49] AlecTaylor: great question.. answered and +1'd [08:49] hey guys anyone here workign with virt-manager im having some really annoying issues connecting to a remote machine with it [08:49] btw morning AlecTaylor Vivek and SpamapS [08:50] Thanks SpamapS [08:51] if you can call 01:00 morning. :) [08:52] If you call 7:51PM morning :P [08:52] SpamapS: yES [08:53] SpamapS: I am well aware of the fact that it is a collection of packages with not centralized configuration. [08:55] eagles0513875_: Hello. [08:55] Vivek: in that case you really just need to point other machines' rsyslogd at the logging server, and make sure the logging server is configured to accept remote messages. [08:55] SpamapS: Do you have any documentation I can refer to ? [08:59] hello... just had to reboot a server this morning (10.10) http://waste.mandragor.org/linux-memleak.png if anyone got a clue for next time it'be much appreciated :( [09:09] Vivek: just the rsyslogd man pages [09:10] yann__: what makes you think there is a memory leak? [09:11] SpamapS, I have no software running, killed all running software and restarted what I could [09:11] and without any software running, using 5GB+ ram [09:11] !paste [09:11] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [09:12] <_godhelpme> hi [09:13] ok [09:14] SpamapS: Thanks [09:15] SpamapS: you ever worked with virt-manager before? [09:18] * AlecTaylor opened up the discussion over whether I should open-source my SaaS: http://openclosedsource.wikia.com [09:29] eagles0513875_: I use it quite often [09:30] yann__: but what do you mean "using" ? [09:30] yann__: htop is somewhat imprecise, try 'free -m' [09:30] SpamapS: im having nothing but issues with it getting it to connect to remote server http://pastebin.com/BV50Gf7S [09:31] yann__: memory used in buffers and cache is memory used to opportunistically speed up the system. There is no memory leak, its just in use for better purposes when your programs haven't explicitly requested it. [09:31] eagles0513875_: I don't use it like that [09:31] and my friend and i have tested on 10.04 10.10 11.04 and same issue [09:31] how do you use it [09:32] even connecting to my localhost to test i have saame issue [09:32] SpamapS: you using xen as the virt technology of choice? [09:37] eagles0513875_, If it is xen you want to connect, you need to enable the unix server in xend. Unfortunately there are other issues which I have not found solutions for [09:38] that didnt do it [09:38] any ideas as to these errors smb http://pastebin.com/BV50Gf7S [09:40] It looks like when I had tried to connect to xen without the change in xend (one has to restart too)... Need to start the machine to remember what exactly I changed. [09:40] restart what the local hos t [09:40] host [09:40] or the remote server [09:43] remote server [09:44] SpamapS, disk cache is yellow in htop :) and no, it's not "faster", i discovered that because I had 5GB swap [09:44] so I killed everything I could to see what was using so much, and was left with that [09:45] smb: giv eme a moment to reboot [09:45] eagles0513875_, no reboot of server needed [09:45] just restart of xend [09:46] that is done [09:47] smb: now what [09:49] http://waste.mandragor.org/memory-day.png mmmh. [09:49] * smb needs a bit trying to type, search and talk at the same time does not work that well [09:50] no worries smb im here for another 40 min [09:52] eagles0513875_, Ok. :) Well so I think you need to have libvirt-bin installed on the remote xen host, then have /etc/xen/xend-config.sxp have a line like "(xend-unix-server yes)", then sudo service xend restart [09:52] libvirt-bin is started already [09:52] let me check the xend-config [09:54] that didnt work :( [09:55] Hm... [09:55] smb: it seems like the remote server is trying to close the connection before i am successfully logged in is the impression im getting from the msg [09:55] eagles0513875_: no I use kvm [09:56] ok SpamapS [09:56] yann__: what does free -m show? [09:56] It looks like that, but I had the exact same one because there was no service running... [09:56] SpamapS, I ll give it a try next time, had to reboot it [09:56] smb: let me try it on my local host first [09:57] yann__: I also usually run 'ps auxw O r' which shows the biggest process last.. [09:57] clint 3859 4.6 5.9 970108 236588 ? Sl Jan15 13:38 /usr/lib/firefox-9.0.1/firefox [09:57] anyway, time to try and sleep [09:57] SpamapS, I usually use htop with sorting by memory :) have a good night, thanks for helping [09:57] eagles0513875_, I would assume if you can ssh ok, the virtmanager connect should work [09:58] im on the localhost already doesnt work cant seem to establish a connection to the local host [09:58] also its seems the remote host is terminating the connection after specifying my password [09:58] smb: you using ssh keys or passwords only [09:58] eagles0513875_, ssh keys [09:59] smb: is it possible to only use passwords or with virt-manager you must have ssh keys [09:59] eagles0513875_, Should be possible to use both. But let me make sure... [10:00] smb: the virt-manager documentation shows remote connectiosn using ssh keys not passwords at least from what i understood [10:00] yann__: also one thing that has gotten me before is if you are using something configured to use HugePages like mysql or postgres.. that memory will show up as "used" by the kernel as soon as you configure the huge pages region. [10:00] eagles0513875_, works both. just without ssh keys there is a local pw dialoque [10:00] I ll have a look, but I don't think I'm using huge pages :) [10:01] smb: im getting the dialogue [10:01] SpamapS: FYI rsyslog is automatically configured on Orchestra. [10:01] but i cant fathom what is causing me to fail to connect [10:02] eagles0513875_, Just for completeness, you checked that on the remote host libvirtd is running? [10:03] Vivek: oh, cool. :) [10:06] eagles0513875_, And one other thing to check is whether "netstat -a|grep unix|grep xend-sock" has some output === gondoi_ is now known as gondoi === Myrtti_ is now known as Guest79446 [10:08] bk smb :) === Vivek is now known as Guest88234 [10:09] eagles0513875, Mind that I could then connect, but instance creation still fails because bug 914788 and bug 914792 [10:09] Launchpad bug 914788 in libvirt "libvirt expexts qemu-dm in wrong path for xen" [Undecided,New] https://launchpad.net/bugs/914788 [10:09] Launchpad bug 914792 in libvirt "libvirt: Unable to complete install: ''b454ca30-add4-8f72-8093-99c938e87b46''" [Undecided,New] https://launchpad.net/bugs/914792 [10:09] smb: im guessing im experiencing the first bug === Guest79446 is now known as Myrtti [10:10] actually first one isnt [10:10] eagles0513875, No that only was _after_ being able to connect. :) [10:10] ahh ok [10:11] guess its time for a bug to be filed [10:11] 2 bugs [10:11] virt-manger should automatically pull ssh-askpass [10:11] no [10:11] I dont think it is a bug [10:11] I think it is your setup === Adri2000_ is now known as Adri2000 [10:12] ikonia: has to be a bug [10:12] why ? [10:12] tried it on a vm of 10.04 10.10 11.04 [10:12] i have everything there libvirt is installed and running [10:12] but didn't smb say he was connecting fine ? [10:13] ikonia: hes using ssh keys not simple passwords [10:13] right - so why ssh-askpass then, as that's normally for keys [10:13] eagles0513875: how are you trying to open the virutal manager ? [10:13] from the kmenu [10:13] should it be run with sudo from the command line [10:14] eagles0513875: have you looked at what the kmenu launcher is doing ? [10:14] no i havent [10:14] eagles0513875: ok - so again, checking this stuff out before logging a bug [10:14] eagles0513875: have you tried launching it manually to get some more verbose output ? [10:14] check the basics, get info, === schmidtm_ is now known as schmidtm [10:15] eagles0513875, I had changed away the authorized_keys to get a pw prompt and that worked, too [10:16] ahh [10:16] I didn't have a problem when I was using it on ubuntu (I'm not any more but it worked fine when I did) [10:16] ikonia: how can i get verbose output to commandline [10:16] ikonia: it worked fine when i tested it out with kvm [10:17] eagles0513875, I think I am a bit confused reading. Did you try the netstat line on the remote host? [10:17] eagles0513875: just launch it by the command line first, see what it does [10:18] ikonia: not doing anything [10:18] smb: no i havent [10:18] eagles0513875: what are you typing on the command line ? [10:18] eagles0513875, Ok, could you try? Then we would know whether the socket interface is up [10:18] smb: what socket interface should i be looking for [10:19] ikonia: virt-manager [10:19] netstat -a|grep unix|grep xend-sock [10:19] unix 2 [ ACC ] STREAM LISTENING 20492 /var/lib/xend/xend-socket [10:19] hi all [10:21] its there and listening smb [10:22] eagles0513875, Ok, so at least can take that from the list... [10:22] ok [10:24] eagles0513875, So what is running locally and remotely? I only tested 11.10 and 12.04 combinations currently [10:25] 11.10 on both [10:25] ok [10:25] well ill be damned i installed kvm on localhost and that connected just fine [10:28] Well, that has always worked. Just that Xen host server came back only with 11.10 and I am not sure how many really tried out with libvirt and xen [10:28] It is quite manual at least [10:28] :-/ [10:29] * smb hopes to improve things until precise release... [10:29] smb: you work with the virtualization stuff? [10:29] yes [10:30] Will open-sourcing my SaaS solution adversely affect my revenue? - Contribute to Discussion and Notes: http://openclosedsource.wikia.com / http://programmers.stackexchange.com/q/130341 [10:30] eagles0513875: don't you have to make a change to tell it to connect to a xen hypervisor instead of kvm [10:30] ikonia, Yes, that is done when you create a new connection [10:30] i did i was just testing ikonia to see if i would have the same issues i am having with xen but with kvm === Guest88234 is now known as Vivek [11:20] morning o/ [11:51] smb: hey I'm on from web chat at school lol [12:05] ohai lynxman [12:39] ikonia: Looks like you're around now eh? [12:39] I am around yes [12:39] Remember that whole MySQL battle from before? ;) [12:39] hello [12:39] I certainly do [12:39] hi ho :) [12:39] heheh [12:39] you'll laugh at the results. ;) [12:39] one of the more interesting ones [12:39] Psi-Jack: please share [12:39] very interested [12:40] Hehe [12:40] Well, the updated kernel and mysql solved the memory usage was mostly solved, but there was an underlying issue causing most of it to begin with. ;) [12:41] go on...... [12:41] We use VMWare vSphere 4.1u2 in our infrastructure for pretty much most every server in our DC, save for a few. [12:42] This particular VMWare Guest was set to have initially 4GB, then 8GB, then finally 16GB, but no matter how much RAM we gave it at the hypervisor level, someone had set a Memory Limit on it to 2GB. [12:42] really, [12:42] So, yeah, the guest was getting 8GB, but only 2GB of it was allowed to be used, and so it was actually swapping out on the HOST OS side to account for it. [12:43] that makes sense for what we saw that night [12:43] however was that hypervisor limit in place on any other machiens ? [12:43] eg: was it effecting any other guests ? [12:43] No, it was set just for that one that we found. [12:44] I didn't know you could do that, limit it at the hypervisor for just one vm [12:44] Yep. [12:44] VMWare can do that. heh [12:44] Our DBA of all people found it! [12:44] I thought the hypervisor settings where global and then everything bellow that was either group policy based, or guest based [12:44] impressive [12:44] and a nice little feature to be aware of for future use [12:44] thanks for sharing Psi-Jack [12:44] I don't claim at all to be a VMWare expert, I know Xen and KVM much more, but VMWare still is bleh. [12:44] what an annoying issue that turned out to be [12:45] Nooooo sh**! [12:45] Psi-Jack: vm-ware is what you have to use in business, so no point dodging it [12:45] apart from redhat pushing kvm as their cloud platform now [12:45] After that, our CPU loads went back to normal, primary use of CPU load is MINIMAL now, and when it is, it's mostly user load, not system load. [12:45] That's what got me is MOST of the load was system load, which means core processes. [12:46] at least you are sorted now [12:46] Yep. :) [12:46] an interesting one, [12:46] A lot more useful now, and it's funny to see the load levels during prime hours are practically nothing at all anymore. [12:46] We have another DB running similarly in another DC, under Xen, and i see a lot of user load, but barely any system load. [12:47] That database however, has a lot going on. Both our flagship SaaS product, and our CMS hostings. ;) [12:47] oh, did someone specify a resource limit for ram? [12:47] patdk-lap: Yeah. LOL [12:47] ya, your miss that unless you use that a lot [12:47] learned something new as well about vmware sphere :D [12:48] hehe [12:48] Yeah, me too! [12:48] setting resource limits is very nice though [12:48] Don't trust VMWAre! hahaha [12:48] expectially for disk access [12:48] Psi-Jack: how do you guys go about creating the xen guests [12:48] Well, our disk access is tied to an FC SAN with FC Disks. ;) [12:48] eagles0513875: We don't. Our Xen stuff is run in Rackspace, they run Xen. heh [12:49] got it [12:49] I [12:49] I don't do Xen anymore myself. Has too many annoying issues like easily corruptable vNIC's. [12:49] I'm thinking I should just run xen inside vmware [12:49] for the last legacy things [12:50] Yikes! [12:50] why yikes? [12:51] Psi-Jack: how are you creating the vNIC's [12:51] you need vsphere 5.0 though [12:51] New bug: #917134 in drbd8 (main) "dbrd8 kernel module and padlock-sha kernel module in deadlock" [Undecided,New] https://launchpad.net/bugs/917134 [12:51] I'm using a bridged networking setup which is very easy to specify and setup the bridge via the network interfaces file [12:51] ikonia: i enabled debugging in libvirt smb is helping me go through that to find out the possible cause of my issue [12:52] eagles0513875: They were fully paravirtualized guests, and just using LVS on them for network directing was causing failures that resulted in mysterious packet loss. [12:52] I'm using my own custom network xen config scripts [12:52] Any kind of advanced routing or packet mangling caused these issues on Xen for me. [12:52] ahh [12:52] well atm libvirt is giving me hell [12:52] Psi-Jack, odd, I didn't have that issue [12:52] And that was Citrix Xen, specifically. [12:52] I also rarely ran it at high load levels though [12:52] The "Commercial Grade" stuff. [12:52] Psi-Jack: it was always one of the drawbacks with xen [12:53] and why cytrix wrote their own network patches for it [12:53] can anyone suggest a good ubuntu load balance app through personal experience ? [12:53] i need to load balance two smtp servers [12:54] lvs [12:54] ipvsadmin [12:54] adm sorry [12:54] simple and easy [12:56] * looking [12:56] thanks very much [12:58] do i need a dedicated server for lvs, or can i run it with other services on a server ? [12:58] You should do it dedicated. [12:58] Or run it along side with the mailservers. [12:59] you can if you are strapped for hardware run the lvs process on the actual service giving servers [12:59] you need to be a little more tidy/thought out, but you could do it if you where strapped for hardware [12:59] i have another server running mysql, doing nothing... i could setup lvs on that server which will balance to the 2 mail servers ? [13:00] You shouldn't run LVS on a database. [13:00] yeah, mail = front end, database = backend, balancing a front end service would put the database at the front [13:00] You would be better off running pacemaker+ldirectord on the two mail servers. [13:00] plus the performance risk [13:00] Just do NOT run a firewall on the same host as LVS [13:01] Psi-Jack: Why not? [13:01] It will eat data for lunch. ;) [13:02] Heck, patdk-lap and Omache taught me that. ;) [13:02] ok, but in essence, just to test, i can use an existing server, load lvs and see how it works - then move it to a dedicated box later [13:02] RootChaos: You should run it on a front-end facing server. Not a database server. [13:03] ok cool [13:03] i can do that [13:03] Often times, people run LVS servers on their webservers. [13:04] I'm one that runs two dedicated VMs just to LVS and does nothing else, but nginx name-based and ip-based proxying. ;) === hito_jp0 is now known as hito_jp [13:05] Psi-Jack: I really don't see the issue of combining those two functions [13:05] Daviey: around? [13:07] koolhead11: briefly [13:07] Daviey: was writing u a mail [13:07] koolhead11: probably better right now [13:07] kinda swapping tasks. [13:10] What do I need to install to host a dhcp server? [13:10] dhcpd [13:11] E: no installation candidate [13:11] Is there a packet that I can just install with apt-get install? [13:12] Here's a tip. [13:12] aptitude search dhcp [13:12] I got it dhcp3-server [13:12] thanks [13:12] That's the older version. [13:13] isc-dhcp-server is version 4 [13:13] ah [13:14] isc-dhcp-server isn't available :( [13:15] There's also a chance I could be wrong. I'm looking at a Debian server of mine. :p [13:15] ah, well I'm on ubuntu [13:17] webmin picked it up, it's working [13:18] heh. [13:18] Glad I'm not the only nut job that uses webmin. :) [13:18] I do it, though, just to be lazy. I actually know how to go in and fix problems by hand, set them up by hand, etc, ;) [13:19] Yeah, I like working in console, but for a quick configurations like setting up users it's just better [13:20] Primarily I use it for user config, apache, dns and dhcp servers [13:21] Oh heck, you don't even utilize it for much then. [13:21] You're just plum lazy. :) [13:22] I actually use it for the whole cluster effect, change one on one, it replicates that change to the others. [13:26] Ah [13:27] hey! [13:29] simple dumb question from a noob.... can i use a ubuntu server as a front end to redirect urls to different web servers within my network? [13:29] yes [13:29] i know how to do it with apache and virtual hosts in a single machine... [13:29] the question then is... how? [13:30] Jeeves_, the issue with combining what two functions? [13:31] skorv: look at squid and jesred, or apache [13:31] skorv: I'd setup nginx with name-based forwards to different transport endpoints as needed for each hostname you need. But that's not the proper way to do what you're asking. The proper way is to have Split-DNS and have internal DNS point to internal server, external DNS point to external IP. [13:32] patdk-wk: Iptables and LVS [13:32] Psi-Jack: the proper way sounds complecated [13:32] iptables and lvs don't mix [13:32] skorv: It's not. [13:32] It is more maintenance, but proper. [13:32] lvs injects packets back in, in the middle of netfilter [13:32] And you don't have a SPF [13:32] lvs packets don't match conntrack ever [13:33] Nope. [13:33] LVS magically moves traffic around without ever telling conntrack about it. ;) [13:33] i'm managing virtual servers as well as physical ones... [13:33]