| --- | Log | opened Sun Mar 20 00:00:44 2005 |
| 00:12 | --- | ---> ijuz__ [~ijuz@p54ABC591.dip.t-dialin.net] has joined #xen |
| 00:20 | --- | <<-- ijuz_ [~ijuz@p54ABBDF3.dip.t-dialin.net] has quit (Ping timeout: 480 seconds) |
| 00:27 | --- | ---> enum [~enum@cpe-24-175-6-236.houston.res.rr.com] has joined #xen |
| 00:27 | enum | yo.. I seem to be having an issue booting using a file based root_fs |
| 00:27 | enum | I set it like the docs say.. ['file:/home/xen/root_fs,sda1,w'] |
| 00:27 | enum | everything starts to boot, and it dies with a kernel panic.. |
| 00:28 | enum | telling me to append a correct root= option |
| 00:28 | enum | do I need to create a dummy device in the root_fs? |
| 00:47 | enum | meh |
| 00:54 | @cw | enum: what is the panic? |
| 00:55 | enum | one sec |
| 00:55 | enum | I'm getting this now |
| 00:55 | enum | Using config file "web.cnf". |
| 00:55 | enum | Error: vbd: Device not found: sda1 |
| 00:56 | enum | i got the panic when I pointed it to an actual device in dom0 (hd1) |
| 00:56 | enum | err hda1 |
| 00:56 | @cw | using the above config. suntax? |
| 00:57 | enum | this is the disk, and root lines |
| 00:57 | enum | disk = ['file:/home/xen/root_fs,sda1,w'] |
| 00:57 | enum | root = "/dev/sda1 ro" |
| 00:58 | enum | basically this is what I did |
| 00:58 | @cw | and when that domain boots it says what? |
| 00:58 | enum | well if I leave it at sda1 it doesnt boot |
| 00:58 | enum | it dies with that error |
| 00:58 | enum | but |
| 00:58 | enum | if I change sda1 to hda1 I get |
| 00:58 | @cw | do you have /dev nodes inside root_fs? |
| 00:59 | enum | it appeared to be filled |
| 00:59 | enum | checking now |
| 00:59 | enum | yeah it is there |
| 00:59 | enum | but no sda |
| 00:59 | enum | in the root_fs |
| 01:00 | @cw | how did you make root_fs? |
| 01:00 | enum | following the docs |
| 01:00 | enum | except I used a fresh stage bz2 file from gentoo.org |
| 01:01 | @cw | gentoo... hmm, not sure about gentoo root filesystems |
| 01:01 | enum | well let me ask this |
| 01:01 | enum | the sda1 in the config |
| 01:01 | enum | where is that supposed to come from? |
| 01:02 | enum | since I'm not using a hard disk for the dom |
| 01:04 | @cw | the block driver in domU will thunk those bits through to the backend in dom0 |
| 01:05 | @cw | so access /dev/sda1 in domU will access root_fs from dom0 |
| 01:06 | enum | hmm |
| 01:06 | enum | okay.. so what would keep it from loading finding sda1 before event running the domU kernel? |
| 01:07 | @cw | is that what is going on? |
| 01:07 | @cw | i thought the error was coming from domU? |
| 01:07 | enum | well if I change sda1 to hda1 |
| 01:07 | enum | in the config.. |
| 01:08 | enum | cw, here is my config --> http://pastebin.com/260495 |
| 01:09 | enum | and this is what I get when I try to boot it |
| 01:09 | enum | vhost_dev xen # xm create web.cnf -c |
| 01:09 | enum | Using config file "web.cnf". |
| 01:09 | enum | Error: vbd: Device not found: sda1 |
| 01:12 | enum | would it be looking for sda1 in dom0 ? |
| 01:13 | enum | and if so.. why would it need to do so |
| 01:22 | @cw | no |
| 01:24 | enum | hmm |
| 01:24 | enum | Im confused |
| 01:25 | enum | so you say that sda1 doesnt have to exist in dom0, and that the domU translates root_fs to sda1.. then why would it say it couldnt find it |
| 01:27 | @cw | no |
| 01:27 | @cw | dom0 will export root_fs to domU as /dev/sda1 |
| 01:28 | @cw | so domU's access to sda1 will be passed to dom0 (via xen) and it will IO to and from the file |
| 01:29 | enum | so.. sda1 in domU technically is the root_fs file in dom0 |
| 01:30 | @cw | sorta |
| 01:30 | --- | ---> muli_ [~muli@nesher3.haifa.il.ibm.com] has joined #xen |
| 01:30 | enum | then why would I get a device not found error |
| 01:30 | @cw | access to /dev/sda1 in domU will be passed by the xen frontend block device to xen which will pass them to the backend device in dom0 which will think them to the file |
| 01:31 | enum | so dom0 has to have scsi support to support sda1 |
| 01:31 | @cw | which domain is giving that error? |
| 01:31 | enum | I beleive dom0 |
| 01:31 | @cw | hmm... |
| 01:31 | enum | this is the dom0 console output |
| 01:31 | enum | vhost_dev xen # xm create web.cnf -c |
| 01:31 | enum | Using config file "web.cnf". |
| 01:31 | enum | Error: vbd: Device not found: sda1 |
| 01:31 | enum | vhost_dev xen # |
| 01:32 | @cw | that message comes from xm? |
| 01:32 | enum | yes |
| 01:33 | enum | but check this out |
| 01:33 | enum | I just did ln -s /dev/hda1 /dev/sda1 then xm create web.cnf -c |
| 01:33 | enum | and it started the domU with a kernel panic |
| 01:33 | enum | so sda1 does have to exist in dom0 ? |
| 01:34 | @cw | what docs are you reading? |
| 01:34 | @cw | got a URL handy? |
| 01:34 | enum | http://www.cl.cam.ac.uk/Research/SRG/netos/xen/readmes/user/user.html |
| 01:38 | @cw | ok, i suspect xm wants to stat /dev/sda1 in dom0 to get major/minor numbers |
| 01:38 | @cw | are you using udev or devfs? |
| 01:38 | enum | devfs |
| 01:38 | @cw | aaaaaaaaaaaaaaaaaaaaarrrrrrrrrrrrrrgggggggggggggggggggh! |
| 01:39 | enum | problem? |
| 01:39 | @cw | gentoo too right? |
| 01:39 | enum | yeah.. I can disable devfs if I have to |
| 01:39 | enum | should I be using udev? |
| 01:40 | @cw | just make a sane device node for /dev/sda1 and see if that fixes it |
| 01:40 | enum | sane? |
| 01:41 | schweeb | sane meaning legitimate |
| 01:42 | schweeb | don't just pull the major/minors out of your ass |
| 01:42 | enum | never had to do that before |
| 01:42 | schweeb | man mknod |
| 01:43 | schweeb | with the assistance of the major/minor reference from the kernel source |
| 01:43 | @cw | MAKEDEV can also do it for you |
| 01:45 | enum | k, that allowed it to boot, but I get the kernel panic |
| 01:45 | enum | VFS: Unable to mount root fs via NFS, trying floppy. |
| 01:45 | enum | VFS: Cannot open root device "sda1" or unknown-block(2,0) |
| 01:45 | enum | Please append a correct "root=" boot option |
| 01:45 | enum | Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(2,0) |
| 01:45 | enum | dont know why it is attempting nfs |
| 01:46 | @cw | does root_fs have /dev/sda1 in it? |
| 01:47 | enum | yes |
| 01:47 | @cw | shit, just change sda1 to something like hda1 you have in both places if need be |
| 01:47 | @cw | 2,0 isn't sda, it's a floppy i think |
| 01:49 | enum | fstab seems to be okay |
| 01:49 | @cw | fstab isn't used that early |
| 01:49 | @cw | what does domU claim it's command line is? |
| 01:50 | enum | Kernel command line: ip=172.16.20.100:1.2.3.4::::eth0:off root=/dev/sda1 ro |
| 01:51 | @cw | hmmm |
| 01:52 | @cw | i would have to check if that's correct but at a glance i wonder why your getting root=/dev/<foo> and not root=801 or similar |
| 01:52 | enum | I create the node by "mknod /dev/sda1 b 8 1" |
| 01:53 | @cw | what does 'ls -l /dev/sda1' say for dom0? |
| 01:53 | enum | vhost_dev xen # ls -l /dev/sda1 |
| 01:53 | enum | brw-r--r-- 1 root root 8, 2 Mar 20 00:46 /dev/sda1 |
| 01:54 | @cw | 8,2? is that right? |
| 01:55 | @cw | 8,1 for sda1 |
| 01:55 | enum | hmm |
| 01:55 | enum | okay.. that was a type-o on my part... |
| 01:55 | enum | kernel cli is now |
| 01:56 | enum | Kernel command line: ip=172.16.20.100:1.2.3.4::::eth0:off root=/dev/sda1 ro |
| 01:56 | enum | but still panic |
| 01:56 | enum | Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(8,1) |
| 01:56 | @cw | what is the *full* kernel boot log? |
| 01:57 | enum | you want me to pastebin it? |
| 01:57 | @cw | sure |
| 01:57 | enum | http://pastebin.com/260501 |
| 01:58 | @cw | where did you get the kernel from? |
| 01:58 | @cw | oh, you made it |
| 01:58 | @cw | the .config it hosed |
| 01:58 | @cw | is hosed |
| 01:58 | enum | wha? |
| 01:58 | enum | I compiled that kernel |
| 01:58 | @cw | your .config you used for your kernel isn't right at a guess |
| 01:59 | enum | hmmm |
| 01:59 | @cw | where did you get the .config from? |
| 02:00 | enum | from the kernel after I patched it |
| 02:00 | enum | I ran "make menuconfig" |
| 02:00 | enum | didnt know there was aplace to get configs from |
| 02:00 | enum | what makes you say it is hosed? |
| 02:01 | @cw | the is a suitable domU config shipped with xen, use that |
| 02:01 | @cw | you've turned on crap you dont need/want and off things you do want |
| 02:01 | enum | hmm |
| 02:01 | enum | let me try another compile |
| 02:01 | @cw | get rid of devfs for one thing |
| 02:01 | enum | udev? |
| 02:01 | @cw | im pretty sure the default xen stuff won't be using that |
| 02:01 | enum | or no |
| 02:01 | @cw | static nodes are fine |
| 02:02 | enum | k |
| 02:02 | enum | back to the drawing board! |
| 02:02 | @cw | eh? |
| 02:02 | @cw | static nodes are the easiest thing to do here |
| 02:02 | @cw | simpler is usually better |
| 02:02 | enum | no devfs on dom0 right? |
| 02:03 | @cw | well, doesnt matter there so much right now |
| 02:03 | @cw | but ideally i wouldn't use it |
| 02:03 | enum | k, Im going to recompile the domU only for now, and see what happens |
| 02:03 | @cw | i wouldn't use devfs *anywhere* *ever* if it was me, but some people like it |
| 02:03 | @cw | yeah, fix domU |
| 02:17 | @cw | enum: you can probably also add devfs=nomount or whatever magic to prevent that from screwing /dev/ up for you as a test |
| 02:19 | enum | hmm nope.. recompile gives me the same kernel panic |
| 02:22 | enum | my config --> http://pastebin.com/260504 |
| 02:22 | enum | oops.. thats the dom0 config |
| 02:23 | @cw | do you have the xen frontend block device? |
| 02:24 | @cw | you have all sorts of shit in this .config |
| 02:24 | @cw | can you not use something SIMPLE like the one shipped with xen? |
| 02:24 | @cw | why aer you turning on lots of weird options? |
| 02:25 | enum | Im redoing this now.. I will have a couple questions once I get to the kernel. should be about 5 minutes or so ;) |
| 02:42 | enum | okay.. so I have the xen package.. where is the configs you are talking about? |
| 02:43 | @cw | in buildconfigs probably |
| 03:31 | --- | <<-- Tv [~Tv@hq.inoi.fi] has quit (Ping timeout: 480 seconds) |
| 03:32 | --- | ---> Tv [~Tv@hq.inoi.fi] has joined #xen |
| 04:09 | enum | yo cw in the buildconfigs dir I see mk.linux-2.6-xenU |
| 04:10 | enum | but no idea how to use it.. and the docs dont say either |
| 04:10 | @cw | thnik it will grab it from there if needed, otherwise cp to .config |
| 04:11 | enum | it is a script |
| 04:11 | enum | http://www.cl.cam.ac.uk/~sos22/replay.bk/buildconfigs/mk.linux-2.6-xenU |
| 04:12 | @cw | ok, let me look |
| 04:13 | enum | I tried using sh ./ and perl to exec it.. but they all fail |
| 04:13 | --- | <<-- muli [~muli@alhambra.mulix.org] has quit (Ping timeout: 480 seconds) |
| 04:13 | @cw | arch/xen/defconfig-xenU |
| 04:13 | @cw | use that |
| 04:38 | enum | cw you sir, are awesome |
| 04:39 | enum | worked jst fine |
| 04:39 | enum | I reinstalled gentoo disabling devfs, and recompiling the kernels |
| 05:00 | @cw | enum: dom0 is mostly OK as-is |
| 05:00 | @cw | enum: but in general i would suggest avoiding devfs where possible |
| 05:01 | @cw | and gentoo :-P |
| 05:01 | enum | gentoo has always been a good setup for me |
| 05:01 | enum | it is all in how you do it ;) |
| 05:03 | @cw | whatever works, im partly trolling, and partly biased :) |
| 05:05 | enum | heh |
| 05:05 | enum | I usually do lfs also.. but I want to streamline my lfs to be a complete slim down for xen |
| 05:07 | @cw | whatever works for you, i personally dont use gentoo myself |
| 05:12 | --- | User: *** cw is now known as anticw |
| 05:16 | --- | User: *** ijuz__ is now known as ijuz |
| 05:17 | enum | yo anticw when starting a new dom does it use memory from dom0? |
| 05:18 | enum | so if I say dom0 has 131027.. and I start a domU with 64M does that come from the dom0s' allocation of 131027? like ( 128 - 64 ) |
| 05:18 | enum | I started 3 domUs' with 64M of ram, and dom0 has 2M left free |
| 05:19 | --- | User: *** anticw is now known as cw |
| 05:19 | @cw | enum: sorry, about to crash (hence then nick change) |
| 05:19 | @cw | enum: no, memory isn't taken from another domain |
| 05:20 | @cw | so if you want to have say 4 domains with 512MB each you need over 2GB of ram (xen uses a little bit itself) |
| 05:20 | enum | hmm.. trying to figure out why the mem in dom0 dropped like that after starting the domUs' |
| 05:20 | @cw | dropped how |
| 05:20 | @cw | ? |
| 05:20 | enum | whent from 128 to 2 |
| 05:20 | enum | freemem |
| 05:20 | @cw | what did? |
| 05:20 | @cw | show me |
| 05:20 | enum | MemTotal: 125920 kB |
| 05:20 | enum | MemFree: 2396 kB |
| 05:20 | enum | thats from dom0 |
| 05:21 | @cw | that's fine |
| 05:21 | enum | very slow |
| 05:21 | @cw | what does Cached say? |
| 05:21 | enum | Cached: 98140 kB |
| 05:22 | @cw | ok, so ~100MB is cached (ie. freeable if apps need it otherwise caching the disk) and 30MB used |
| 05:22 | enum | hmm |
| 05:22 | enum | okay |
| 05:22 | @cw | what is slow? |
| 05:23 | enum | any commands |
| 05:23 | enum | even in the shell I will press the up arrow to get the last command |
| 05:23 | @cw | in dom0 or domU? |
| 05:23 | enum | takes about 2 seconds |
| 05:23 | enum | dom0 |
| 05:23 | @cw | is it swapping? |
| 05:23 | enum | nope |
| 05:23 | @cw | yeah, it wouldn't be with Cached that high |
| 05:24 | @cw | slow before domU was started? |
| 05:24 | enum | not at all |
| 05:24 | @cw | odd |
| 05:24 | enum | and this is an AMD64 with 2 GB Ram |
| 05:24 | @cw | is domU doing anything? |
| 05:24 | enum | nope |
| 05:25 | @cw | "xm list" says? |
| 05:25 | enum | Name Id Mem(MB) CPU State Time(s) Console |
| 05:25 | enum | Domain-0 0 123 0 r---- 1501.9 |
| 05:25 | enum | ox 8 63 0 -b--- 5.1 9608 |
| 05:25 | enum | psql 5 63 0 ----- 306.8 9605 |
| 05:25 | enum | web 6 63 0 ----- 243.8 9606 |
| 05:26 | @cw | and again? |
| 05:26 | enum | Name Id Mem(MB) CPU State Time(s) Console |
| 05:26 | enum | Domain-0 0 123 0 r---- 1508.4 |
| 05:26 | enum | ox 8 63 0 -b--- 5.1 9608 |
| 05:26 | enum | psql 5 63 0 ----- 325.4 9605 |
| 05:26 | enum | web 6 63 0 ----- 276.2 9606 |
| 05:27 | @cw | so three domains are using the cpu right now |
| 05:27 | enum | yes |
| 05:27 | @cw | is there a lot running? |
| 05:28 | enum | ive got one compiling apache, and the other running a report |
| 05:28 | @cw | do xm help and look at atropos |
| 05:28 | @cw | i have to get some sleep, but that should let you adjust the sched. to suit |
| 05:28 | @cw | well, somewhat |
| 05:28 | @cw | gotta go now |
| 05:28 | --- | User: *** cw is now known as anticw |
| 05:29 | enum | thanks cw.. I really appreciate it |
| 05:47 | --- | <<-- deac [~deac@xdsl-195-14-216-202.netcologne.de] has quit (Ping timeout: 480 seconds) |
| 06:01 | --- | ---> deac [~deac@xdsl-81-173-138-211.netcologne.de] has joined #xen |
| 06:24 | --- | <<-- monrad [~monrad@213083190130.sonofon.dk] has quit (Ping timeout: 480 seconds) |
| 07:37 | --- | <<-- enum [~enum@cpe-24-175-6-236.houston.res.rr.com] has quit (Ping timeout: 480 seconds) |
| 08:11 | --- | <<-- muli_ [~muli@nesher3.haifa.il.ibm.com] has quit (Quit: My damn controlling terminal disappeared!) |
| 08:20 | --- | <<-- visik7 [~ciao@host155-36.pool80182.interbusiness.it] has quit (Read error: Connection reset by peer) |
| 08:39 | --- | ---> muli [muli@alhambra.mulix.org] has joined #xen |
| 10:25 | --- | ---> monrad [~monrad@213083190130.sonofon.dk] has joined #xen |
| 10:40 | --- | ---> Nigelenki [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has joined #xen |
| 10:45 | --- | <<-- Bluefox [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has quit (Ping timeout: 480 seconds) |
| 11:28 | --- | ---> visik7 [~ciao@host155-36.pool80182.interbusiness.it] has joined #xen |
| 11:28 | visik7 | hi |
| 11:29 | visik7 | does fb work normally on dom0 ? |
| 11:34 | @Sir_Ahzz | as logn as no other domain needs to use it. |
| 11:34 | @Sir_Ahzz | it's just vga hardware. |
| 11:38 | visik7 | why don't you use a diff/patch system instead of mkbuildtree script ^? |
| 11:41 | @Sir_Ahzz | because the code is nearly identical for the different kernel trees. |
| 11:41 | @Sir_Ahzz | reduces replication of files/source. |
| 11:42 | @Sir_Ahzz | it's also easier for the developers to test with. |
| 11:45 | visik7 | understood |
| 12:45 | --- | ---> Icy [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has joined #xen |
| 12:45 | --- | <<-- Nigelenki [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has quit (Read error: Connection reset by peer) |
| 13:12 | --- | <<-- visik7 [~ciao@host155-36.pool80182.interbusiness.it] has quit (Read error: Connection reset by peer) |
| 13:53 | --- | <<-- monrad [~monrad@213083190130.sonofon.dk] has quit (Read error: Operation timed out) |
| 15:22 | --- | ---> Nigelenki [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has joined #xen |
| 15:22 | --- | <<-- Icy [~bluefox@pcp484971pcs.whtmrs01.md.comcast.net] has quit (Read error: Connection reset by peer) |
| 15:34 | --- | ---> monrad [~monrad@0x535b06c0.ronxx3.adsl-dhcp.tele.dk] has joined #xen |
| 15:39 | sunny | 10 |
| 15:39 | sunny | err, sorry |
| 16:23 | --- | <<-- monrad [~monrad@0x535b06c0.ronxx3.adsl-dhcp.tele.dk] has quit (Ping timeout: 480 seconds) |
| 16:31 | --- | ---> visik7 [~visi@host155-36.pool80182.interbusiness.it] has joined #xen |
| 16:36 | --- | <<-- visik7 [~visi@host155-36.pool80182.interbusiness.it] has quit (Quit: ) |
| 16:36 | --- | ---> visik7 [~visi@host155-36.pool80182.interbusiness.it] has joined #xen |
| 17:03 | muli | n |
| 18:27 | --- | <<-- deac [~deac@xdsl-81-173-138-211.netcologne.de] has quit (Ping timeout: 480 seconds) |
| 18:39 | --- | ---> Lampy [~jon@dynamic-62-56-34-233.park-s46b.dslaccess.co.uk] has joined #xen |
| 18:40 | --- | ---> deac [~deac@xdsl-213-196-207-190.netcologne.de] has joined #xen |
| 18:55 | --- | <--- Lampy [~jon@dynamic-62-56-34-233.park-s46b.dslaccess.co.uk] has left #xen () |
| 18:56 | --- | <<-- visik7 [~visi@host155-36.pool80182.interbusiness.it] has quit (Quit: visik7) |
| 19:02 | --- | ---> visik7 [~visi@host155-36.pool80182.interbusiness.it] has joined #xen |
| 19:35 | --- | <<-- visik7 [~visi@host155-36.pool80182.interbusiness.it] has quit (Remote host closed the connection) |
| 23:44 | --- | <<-- sunny [sunny@opencurve.org] has quit (Quit: test) |
| 23:44 | --- | ---> sunny [sunny@opencurve.org] has joined #xen |
| 23:59 | --- | <--- VS_ChanLog [~stats@ns.theshore.net] has left #xen (Rotating Logs) |
| 23:59 | --- | ---> VS_ChanLog [~stats@ns.theshore.net] has joined #xen |
| --- | Log | closed Mon Mar 21 00:00:16 2005 |
