1362874607 Q * bonbons Quit: Leaving 1362879405 J * BenG ~bengreen@cpc35-aztw23-2-0-cust207.18-1.cable.virginmedia.com 1362879622 Q * BenG 1362882185 J * kolorafa_ ~irc@cfh105.neoplus.adsl.tpnet.pl 1362882539 Q * kolorafa Ping timeout: 480 seconds 1362891664 J * clopez ~clopez@21.102.11.37.dynamic.jazztel.es 1362893012 Q * clopez Read error: Operation timed out 1362893017 M * Bertl off to bed now ... have a good one everyone! 1362893022 N * Bertl Bertl_zZ 1362900079 Q * wmp Remote host closed the connection 1362904215 Q * arekm Read error: No route to host 1362904265 J * arekm ~arekm@ixion.pld-linux.org 1362906111 Q * nkukard_ Read error: Operation timed out 1362906167 J * nkukard_ ~nkukard@196-210-204-27.dynamic.isadsl.co.za 1362907662 J * bonbons ~bonbons@2001:a18:20a:1601:117c:747d:1ed8:2c9 1362911380 J * wmp ~wmp@2001:41d0:1:8616::1 1362918277 Q * ircuser-1 Read error: Operation timed out 1362919374 Q * Aiken Remote host closed the connection 1362919652 J * Seattle ~chatzilla@c-24-34-74-228.hsd1.ma.comcast.net 1362919941 M * Seattle Hi All - am working on creating failover for vservers using drbd/pacemaker. Making progress...have a /Vservers mount for vserver data. Q: is a good approach to create sub folders in each vserver, and link back to /etc/vservers and /var/lib/vservers? I.e., right now the data move over from one server to another, and I have a stab at an LSB in pacemaker, but the failover server knows... 1362919943 M * Seattle ...nothing about specific vservers. 1362920153 M * Seattle Or....as one post suggests, would you actually move the /var/lib/vservers/vservername to /vservers/vservername/barrier/var? 1362920690 M * kolorafa_ what do you mean by failover, active network parition on remote server for vservers? 1362920827 M * kolorafa_ Can i ask about fail scenario? What when your server goes down, then ofc you need to boot up second one, but the your main going up and what with synchronization of full block device? isn't that some kind of overkill? or there is some additional commit/changes log? 1362921139 M * kolorafa_ i understand it that, the barier you set on directory above so for example on /vservers or /var/lib/vservers, the barier is only a simple flag, so you can set multiple directory for "more security", it could only affects your admin experience when using chroot 1362921345 M * Seattle Might be overkill....but want high availability. E.g., if one server goes down, the other kicks in. As of now, working in terms of the vserver data getting passed over only. Are you suggesting that something like rsync would be enough in terms of data replication? Am planning on having multiple vservers....but could rsync the whole /Vservers data area....but still need some way for the... 1362921347 M * Seattle ...failover box to know about each vserver. Would you actually "move" the /var/lib/vservers folder or symlink it? Does anything need to happen in terms of /etc/vservers? 1362921432 M * kolorafa_ but is it safe for filesystem? if you power down server in half write you need to do fsck on FS 1362921449 M * kolorafa_ so on server down you need to run fsck, then start vservers 1362921458 M * kolorafa_ at it will be not so fast 1362921559 M * kolorafa_ rsync is not good solution for large amount of files 1362921616 M * kolorafa_ symlink is symlink, only tells system where to go, and he follows the patch, and barier is need to be set on directory above the real vserverGuest directory 1362921661 M * Seattle OK....cool. If understand, fsck would help potentially get master node back in business....but if had other hardware failure would still like failover path.... 1362921673 N * Bertl_zZ Bertl_oO 1362921684 M * kolorafa_ as for my, the best solution for availability would be separete system for data and another for cpu/running stuff 1362921703 M * kolorafa_ so if one of data server go down, you have second one still working with the same data, like raid1 1362921748 M * Seattle Thanks - understood...am exploring this at the moment though, and would like to see how to get it working.... 1362921755 M * kolorafa_ and if the "working" server go down you will have "website is down" problem if you dont have some loadbalancing before 1362921790 M * Seattle right 1362921800 M * kolorafa_ so the question is what exactly you want to be high availability, some webpages or whole vservers with random apps 1362921809 M * Seattle whole vservers 1362921859 M * kolorafa_ apps can only run on single machine, so if it goes down, there isn't a way to resume them, only start that apps from secondary server 1362921969 M * Seattle Right...but my sense of pacemaker/drbd will do this....but need some dynamic way to tell failover box to start vservers (am trying LSB option in pacemaker right now) as well as 2nd machine needs to know what vservers exit. So I think there needs to be someway to pass over either/both the /etc/vserrvers /var/lib/vservers data... 1362921971 M * kolorafa_ using NetworkFileSystem to some raid system you can be 100% sure about data consistent, but on block? I don't think so, i think fsck will be needed to run on secondary server before you run it 1362922018 M * Seattle Yes - each vserver is in its own logical volume..... 1362922028 M * Seattle ....and it's down drbd 1362922052 J * ircuser-1 ~ircuser-1@35.222-62-69.ftth.swbr.surewest.net 1362922060 M * kolorafa_ I would rather make a central vserver list, so if some physical host goes down you list vservers from that list and run secondary machine from them 1362922184 M * kolorafa_ and yes 1362922197 M * kolorafa_ you can move your vservers directory into your paritions 1362922204 M * Seattle Would think "central vserver list" is on master node...question I think is how dynamically tell failover node what that list is, so that vservers start automagically 1362922259 M * kolorafa_ if you manage large ammount of vservers, you need to be abble to start/stop servers from some kind of panel 1362922280 M * kolorafa_ the same panel should monitor hosts system if there are up or down 1362922325 M * Seattle Thanks Kolorafa - apologize, need to run for bit....will "stay on", but be back in about an hour...appreciate the help 1362922623 M * kolorafa_ vserver drbd partition mounting: /drdb/somevserver/, symlink: /etc/vservers/somevserver -> /drdb/somevserver/config, symlink: /etc/vservers/somevserver/vdir -> /drdb/somevserver/data, it think that barier should be on every /drdb/somevserver and optional /drdb, but then quota will not work and if someone use 100% you would not have space for config files, so ... 1362922717 M * kolorafa_ i would rather make something like you push using for example using rsync /etc/vservers/somevserver directory on every config change, and just symlink /etc/vservers/somevserver/vdir to mounted drdb partition :) 1362925031 J * hijacker_ ~hijacker@cable-84-43-134-121.mnet.bg 1362926224 J * Roomster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362926224 Q * Romster Read error: Connection reset by peer 1362926343 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362926343 Q * Roomster Read error: Connection reset by peer 1362927244 N * Bertl_oO Bertl 1362927253 M * Bertl morning folks! 1362927929 M * Seattle Am back...to reiterate, am experimenting with failover design that includes vservers each in own LV as a backing device for drbd along with pacemaker. So far, get data vserver data to failover succesfully and have a draft LSB in pacemaker config that calls util-vserver. 1362928020 M * Seattle Q: how to tell the failover server about each vserver. Link on vserver site suggests moving var/lib/vserver to /vserver/vservername/barrier/var. Seems though like knowledge of specific vservers is in /etc/vservers....would some sort of symlink work? Can't imagine moving /etc/vservers makes sense... 1362928044 M * Seattle PS am using Debian 6 1362928765 M * Bertl first, /var/lib/vserver(s) is not the canonical place for guest data 1362928789 M * Seattle Right - is it /etc/vservers? 1362928798 M * Bertl /vservers is where guest data is supposed to reside 1362928832 M * Bertl guest config is in /etc/vservers 1362928892 M * Bertl it isn't a good idea to put the guest config (i.e. /etc/vservers/) inside or even on the same partition as /vservers 1362928942 M * Seattle Right...so we're putting the guest data in /Vservers (does case matter) where each one has its own LV, drbd mount, and mount (e.g., /Vserveres/vserverx). If sounds OK, the question I have is how to tell the failover server where the vserver data are....do I need to sync the /etc/vservers/ data somehow? 1362928957 M * Seattle By the way , good morning (or a least it's morning over here)! 1362928994 Q * Romster Read error: No route to host 1362928996 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362929010 M * Jb_boin its only config files, you can asyncroneously copy it between servers it wont be a problem 1362929027 M * Bertl yes, and it should be /vservers not /Vservers :) 1362929032 M * Jb_boin as it shouldnt be modified too often 1362929060 M * Seattle Ah...(on /vserver). And are you thinking something like rsync? 1362929089 M * Seattle I.e., rsync the /etc/vservers data? 1362929159 M * Jb_boin you can use simple crontabed rsync or if you want instantaneous replication of the configurations modifications you can use lsyncd or put the /etc/vservers on drdb too (but it might not be the safest/simplest thing) 1362929192 M * Seattle yes, would hesitate to use drbd with /etc/vservers... 1362929223 M * Jb_boin imo it would be overkill and there are more risks of failure than using a KISS solution 1362929242 M * Seattle OK...thanks! 1362929366 J * Roomster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362929367 Q * Romster Read error: Connection reset by peer 1362929601 Q * hijacker_ Quit: Leaving 1362929976 Q * Roomster Remote host closed the connection 1362930258 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362931073 Q * Romster Ping timeout: 480 seconds 1362933011 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1362933546 M * Guest1048 hi Romster 1362933846 M * Romster hi, i'm having crap power night.. 1362933859 M * Romster so bad even my ups can't keep up with it 1362936651 Q * Seattle Remote host closed the connection 1362940024 M * kolorafa_ Seattle, you can always put everyting in some database or remote server, and make a simple scripts that get and create config from that, so only one thing that is left to know where data are on your network/disk/other :) vserver's are simple as that, few files in /etc/vservers that defines what/where and data files :) 1362942871 J * ksn ~ksn@8ta-150-203-25.telkomadsl.co.za 1362942909 P * ksn 1362943212 J * ksn ~ksn@8ta-150-203-25.telkomadsl.co.za 1362943219 P * ksn 1362943272 J * ksn ~ksn@8ta-150-203-25.telkomadsl.co.za 1362943292 P * ksn 1362943832 N * Guest1048 yang567 1362947317 J * ksn ~ksn@00019501.user.oftc.net 1362948121 J * Aiken ~Aiken@2001:44b8:2168:1000:21f:d0ff:fed6:d63f 1362948437 P * ksn 1362949487 Q * cuba33ci Read error: Connection reset by peer 1362949575 J * cuba33ci ~cuba33ci@114-36-231-164.dynamic.hinet.net 1362950271 J * clopez ~clopez@108.30.165.83.dynamic.mundo-r.com 1362950311 J * hijacker_ ~hijacker@cable-84-43-134-121.mnet.bg 1362953746 J * ksn ~ksn@00019501.user.oftc.net 1362954808 J * AraBMoney ~linsc44@63.143.38.21 1362954812 M * AraBMoney !help 1362955039 Q * ksn Quit: leaving 1362955089 J * Ghislain ~aqueos@adsl1.aqueos.com 1362955238 Q * AraBMoney 1362957970 Q * Ghislain Quit: Leaving. 1362959427 Q * bonbons Quit: Leaving