--- Day changed --- Log opened Sat Jul 30 23:59:02 2005 02:40 -!- commander [commander-@p549268CD.dip.t-dialin.net] has joined #uml 03:07 -!- besonen [~besonen@dsl-db.pacinfo.com] has quit [Read error: Connection reset by peer] 04:11 -!- linbot [~supybot@ns.theshore.net] has quit [Read error: Connection reset by peer] 04:11 -!- linbot` [~supybot@ns.theshore.net] has joined #uml 04:13 -!- pasky [pasky@pasky.noc.xs26.net] has joined #uml 04:31 -!- Newsome [~sorenson@216-190-206-130.customer.csolutions.net] has left #uml [Leaving] 04:40 -!- Tv [~Tv@GMMCMXXXIV.dsl.saunalahti.fi] has joined #uml 04:51 -!- linbot` [~supybot@ns.theshore.net] has quit [Remote host closed the connection] 04:52 -!- linbot [~supybot@ns.theshore.net] has joined #uml 05:39 -!- flatronf700B [~flatronf7@202.75.186.154] has joined #uml 07:50 -!- Nem^ [~Nem@p54ABF1C6.dip.t-dialin.net] has joined #uml 07:50 -!- Nem^^ [~Nem@p54ABF1C6.dip.t-dialin.net] has joined #uml 07:53 -!- shred00 [~brian@linux.interlinx.bc.ca] has left #uml [] 09:24 -!- Electric1lf [~david@snap.swebits.org] has joined #uml 09:26 -!- ElectricElf [~david@electricelf.chair.oftc.net] has quit [Remote host closed the connection] 10:13 -!- Tv [~Tv@GMMCMXXXIV.dsl.saunalahti.fi] has quit [arion.oftc.net neutron.oftc.net] 10:13 -!- lokkju [~lokkju@68-116-117-174.dhcp.astr.or.charter.com] has quit [arion.oftc.net neutron.oftc.net] 10:13 -!- silug [~steve@206.80.72.34] has quit [arion.oftc.net neutron.oftc.net] 10:13 -!- SNy [a7c816ce0c@bmx-chemnitz.de] has quit [arion.oftc.net neutron.oftc.net] 10:13 -!- Nem^^ [~Nem@p54ABF1C6.dip.t-dialin.net] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- Nem^ [~Nem@p54ABF1C6.dip.t-dialin.net] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- flatronf700B [~flatronf7@202.75.186.154] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- commander [commander-@p549268CD.dip.t-dialin.net] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- nextime [~nextime@213-140-6-96.fastres.net] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- fo0bar [fo0bar@cromulent.colobox.com] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- kybe [~kyrre@legolas.iu.hio.no] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- buggs [~noidentd@n80-237-228-135.cnet.hosteurope.de] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- nachoguy [~boster@67.95.66.69] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- Darky [~sven@klecker.debilian.org] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- desaster [desaster@yog-sothoth.dragonlight.fi] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- dgraves [~agraves@inet-netcache2-o.oracle.com] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- tomimo [~kurre@a84-231-4-152.elisa-laajakaista.fi] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- Shaun [ndci@ip68-111-70-41.oc.oc.cox.net] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- zimb0 [~zimbo@callisto.dom.bonis.de] has quit [arion.oftc.net unununium.oftc.net] 10:13 -!- nessie [~nessie@lucifer.nerdfest.org] has quit [arion.oftc.net unununium.oftc.net] 10:14 -!- Nem^^ [~Nem@p54ABF1C6.dip.t-dialin.net] has joined #uml 10:14 -!- Nem^ [~Nem@p54ABF1C6.dip.t-dialin.net] has joined #uml 10:14 -!- flatronf700B [~flatronf7@202.75.186.154] has joined #uml 10:14 -!- commander [commander-@p549268CD.dip.t-dialin.net] has joined #uml 10:14 -!- nextime [~nextime@213-140-6-96.fastres.net] has joined #uml 10:14 -!- zimb0 [~zimbo@callisto.dom.bonis.de] has joined #uml 10:14 -!- nessie [~nessie@lucifer.nerdfest.org] has joined #uml 10:14 -!- Shaun [ndci@ip68-111-70-41.oc.oc.cox.net] has joined #uml 10:14 -!- tomimo [~kurre@a84-231-4-152.elisa-laajakaista.fi] has joined #uml 10:14 -!- dgraves [~agraves@inet-netcache2-o.oracle.com] has joined #uml 10:14 -!- desaster [desaster@yog-sothoth.dragonlight.fi] has joined #uml 10:14 -!- Darky [~sven@klecker.debilian.org] has joined #uml 10:14 -!- nachoguy [~boster@67.95.66.69] has joined #uml 10:14 -!- buggs [~noidentd@n80-237-228-135.cnet.hosteurope.de] has joined #uml 10:14 -!- kybe [~kyrre@legolas.iu.hio.no] has joined #uml 10:14 -!- fo0bar [fo0bar@cromulent.colobox.com] has joined #uml 10:15 -!- Tv [~Tv@GMMCMXXXIV.dsl.saunalahti.fi] has joined #uml 10:15 -!- lokkju [~lokkju@68-116-117-174.dhcp.astr.or.charter.com] has joined #uml 10:15 -!- silug [~steve@206.80.72.34] has joined #uml 10:15 -!- SNy [a7c816ce0c@bmx-chemnitz.de] has joined #uml 10:16 -!- Newsome [~sorenson@216-190-206-130.customer.csolutions.net] has joined #uml 10:39 -!- shred00 [~brian@linux.interlinx.bc.ca] has joined #uml 10:39 < shred00> is CONFIG_TTY_LOG gone in 2.6.10 UML? 12:04 -!- ElectricElf [~david@electricelf.chair.oftc.net] has joined #uml 12:06 -!- besonen [~besonen@dsl-db.pacinfo.com] has joined #uml 12:19 -!- Blaisorbl [~blaisorbl@host157-148.pool8289.interbusiness.it] has joined #uml 12:19 -!- Blaisorbl is now known as Blaisorblade 12:24 -!- Electric1lf [~david@snap.swebits.org] has quit [Ping timeout: 1200 seconds] 12:24 -!- lokkju [~lokkju@68-116-117-174.dhcp.astr.or.charter.com] has quit [Quit: Leaving] 12:26 < shred00> i want to start a uml so that it a) logs console output somewhere and b) i can "connect" to the console using a "predeterminable" access hook. having to get a pty out of the boot log does not work well for automation. any ideas? 12:28 < Blaisorblade> Using screen? 12:28 < Blaisorblade> Also, if you have old PTYs enabled, you could probably use the tty/pty= option. 12:29 < Blaisorblade> For screen, it's something like: screen "uml-command" and then detach, or screen to create detached. 12:30 < Blaisorblade> On the same host, you can then reattach to the session. 12:31 < shred00> Blaisorblade: "old ptys" seems most viable. so i give uml "con=/dev/pty1" (for example) then? 12:31 < Blaisorblade> con1=pty:/dev/pty1, more likely.... 12:32 < Blaisorblade> to attach to the other side, you can use minicom or screen... 12:33 < Blaisorblade> On the I/O page there's something about this.... 12:34 < Blaisorblade> Looking on this... 12:34 < shred00> don't pty's have a master/slave relationship where once side opens the master and the other side opens a slave or something like that? 12:35 < Blaisorblade> exactly, I was looking for that... 12:35 < Blaisorblade> : /dev/ptypN seems to be the master side, /dev/ttypN the slave side... 12:37 < Blaisorblade> The master side is first opened by UML, the slave side by minicom. 12:37 < Blaisorblade> Look at the end of: http://user-mode-linux.sourceforge.net/input.html 12:38 < shred00> yeah, on fc3/udev they are conveniently /dev/pty/{m,s}[0-9] 12:38 < Blaisorblade> Yes, that's the DevFS naming. In that example, you must follow the instructions for the first UML (he uses ssl1=pty, you'd use con1=tty:/dev/pty/m0) 12:39 < Blaisorblade> And then, instead of passing the slave side to UML and running minicom inside UML, you'd run minicom directly on the host. 12:39 < Blaisorblade> Attached on the slave side. 12:39 < Blaisorblade> In case of problems, you could try swapping the side, but it seems this is the correct setup. 12:40 < Blaisorblade> Remember, on new host kernels, that legacy PTYs are an option... 12:40 < Blaisorblade> CONFIG_LEGACY_PTYS 12:41 < shred00> damn. not set. 12:41 < shred00> do even old style ptys allow you to open the individual master devices (i.e. /dev/pty/m0) rather than the general master interface (/dev/pty/m)? 12:43 < Blaisorblade> ? I chose old style exactly because only they allow this. With *new* pty's you open the general master device (normally /dev/ptmx, I think this is the case even with udev) and figure which is the guest with some ioctl 12:44 < Blaisorblade> That is done inside glibc - there are higher-level API for this, but I think this is how things work. 12:44 < shred00> Blaisorblade: ahhh. ok then. let me enable those and give that a try. 12:45 < shred00> essentially i am trying emulate serial console term servers and daemons that connect to serial consoles via a tcp port, and the accompanying daemon that does such and logs all console i/o to a file. 13:10 < connelly> shred00: for automation, I use uml_mconsole 13:10 < connelly> in the linux command line, set the umid=some_unique_identifier_for_the_Guest 13:10 < shred00> connelly: no, i want to be able to log console output. unless there is some uml_mconsole usage that i am missing. 13:11 < connelly> tty=`uml_mconsole $boxname config con1 | cut -d: -f2` 13:11 -!- loko [~rbrown@c-67-163-203-163.hsd1.pa.comcast.net] has quit [Ping timeout: 480 seconds] 13:11 < connelly> screen $tty 13:12 < connelly> and when you launch the guest OS, do it inside a screen session with con0=fd:0,fd:1 13:12 < connelly> and make sure screen has a huge default buffer size 13:12 < connelly> and you should be great 13:14 < Blaisorblade> Hmm, using uml_mconsole like that is nice for automation! 13:14 < shred00> ahhhh. now something along those lines might just work! thanks! 13:18 < connelly> if only the uml_mconsole config option let me grab the IP assigned to various interfaces.. 13:20 < shred00> i have my uml session starter give the uml an ip on the cmdline and then have a script in the uml that extracts if from /proc/cmdline and sets it. 13:21 < Blaisorblade> Exactly, I have that too, and actually it was took from the scripts created by UMLbuilder 13:22 < connelly> can you give it an IP with eth0=daemon? I thought you could only give it MAC addresses? 13:22 < Blaisorblade> UMLbuilder was a root_fs image creator. 13:22 < Blaisorblade> connelly: you can't give an IP with any transport. 13:23 < connelly> didn't think so 13:23 < Blaisorblade> I put something like ETH0_IP_SET=, and then have a script parsing that. And that's what shred00 does too. 13:23 < connelly> cool 13:23 < connelly> okie, will do that 13:23 < Blaisorblade> connelly: with TUNTAP, on the cmd line you pass the IP of the host side, not of the UML side. 13:23 < connelly> I had been just setting the MAC address and using DHCP, but that's nicer 13:23 < connelly> Blaisorblade: I see 13:24 < shred00> dhcp would work. you could set static leases by mac addr. but it's just too much overhead. 13:24 < connelly> do you guys reccommend any utilities to do filesystem diffs between COWs? 13:25 < connelly> shred00: indeed, it is, which is why I was asking about alternatives 13:25 < Blaisorblade> Not known any... except the obvious way (mounting both COWs and diffing the contents). 13:25 < connelly> yeah.. using diff (even in recursive binary quiet mode) isn't ideal 13:26 < connelly> because if an entire directory is unique to one filesystem, it will just mention it 13:26 < connelly> I was thinking of using mtree or something 13:26 < connelly> or possibly play with UNIONFS 13:26 < connelly> unionfs on top of a COW just seemed a bit redundant in some ways 13:27 < connelly> but the problem is that COWs diverge over time due to things like software updates 13:27 < shred00> even worse, cows replicate each other due to things like software updates 13:27 < connelly> if I have two COWs that share the same original image, that have both done things like upgrade thier OSes, I'd like to be able to create a new common base file 13:28 < connelly> exactly 13:28 < shred00> you want to essentially do a block-for-block comparison of cows and if a modified block is the same in all cows, push it down to the base and empty the block in cows 13:29 < connelly> also, is it possible to chain COWs? so you can use them (at a performance hit, obviously) as snapshots? 13:29 < connelly> shred00: yes, but at the filesystem, not block level 13:29 < shred00> connelly: no, you want to do it at the block level. cows are just block level "layers" 13:29 < connelly> because if you upgrade the OS on two COWS in the same way on two guests, that doesn't guaruntee that the same blocks would be modified 13:30 < connelly> even if you change the same files in the same way 13:30 < shred00> connelly: hrm. yeah, it would have to be done in the context of the filesystem. 13:31 < connelly> I'm thinking of something more that you find all of the files that have the same metadata and MD5 hashes and take that UNION as a tar and use that as the new base file and the differences from that for the two new COW files 13:33 -!- Blaisorblade [~blaisorbl@host157-148.pool8289.interbusiness.it] has quit [Ping timeout: 480 seconds] 13:33 < shred00> connelly: and your environment allows for you to shut down all of the UMLs to perform this operation? 13:36 -!- Blaisorblade [~blaisorbl@host157-148.pool8289.interbusiness.it] has joined #uml 13:42 < shred00> when i use con=pts when starting a uml and then screen to connect to the console and then shutdown the uml, it goes to "Halting system..." in the screen session fine but the uml never seems to actually stop (i,e. exiting on the terminal it was started on). is this normal? 13:42 < connelly> shred00: uml_mconsole $boxname exec shutdown -h now 13:44 < shred00> hrm. my uml_mconsole does not have "exec". but i am presuming this just executes a command on the uml, as it it were entered in a root shell? 13:44 < connelly> I use bs5 13:44 < connelly> yes 13:45 < shred00> so that would not really produce results different from the same command on a logged in console then? 13:46 < shred00> anyone had any luck with the port: input option? 13:46 < Blaisorblade> shred00: exactly. 13:46 < Blaisorblade> connelly: nice seeing people using exec. 13:46 < Blaisorblade> exec->mconsole_exec 13:47 < shred00> so i would still have this problem that the uml shuts down but does not exit. 13:47 < Blaisorblade> Well, if you have telnet installed, it should work... 13:47 < Blaisorblade> sorry, telnetd. 13:47 < shred00> telnetd? it requires telnetd? that might be my problem then. 13:49 -!- commander [commander-@p549268CD.dip.t-dialin.net] has quit [Quit: Segmentation fault] 13:49 < shred00> hrm. installed telnetd. still no joy. does telnetd have to be enabled, listening? i would have thought the uml process would have just opened the port directly for connection by a telnet client 13:51 < Blaisorblade> No, it mustn't be enabled... it must just be located in /usr/sbin/in.telnetd, and /usr/lib/uml/port-helper must be there. 13:51 < Blaisorblade> And the second thing doesn't quite happen on some 64-bit distros. 13:53 < connelly> I mostly use SSH to communicate with my guest OSess 13:53 < shred00> connelly: yeah, me too. but i need to be able to "monitor" console output and reach them if their networking gets borked. 13:54 < connelly> agreed-which is why I use screen $tty occasionally 13:54 -!- Tv [~Tv@GMMCMXXXIV.dsl.saunalahti.fi] has quit [Quit: Client exiting] 13:55 < shred00> ahhh. closer. 13:55 < shred00> telnetd: Failed to chown slave side of pty: Operation not permitted 13:55 < shred00> . 13:55 < shred00> telnetd: Slave failed to initialize. 13:55 < connelly> is there a way to mount COW'ed files via loopback to a host OS? 13:56 < shred00> ah ha. have to start uml as root in order to use the port:9001 option. :-( 13:57 < shred00> but at least con=port:9001 shuts down cleanly 14:04 < Blaisorblade> shred: I think that's strange: it's reasonable that telnetd requires root privileges to be started (but maybe it's setuid somewhere), but UML probably closes indeed the terminal already... maybe it's just not caught by the terminal emulator. 14:04 < Blaisorblade> When I do "screen uml-startup.sh", which uses con0=fd:0,fd:1, it works fine even on exit. 14:05 < shred00> all con= modes seem to shutdown cleanly except pts. 14:05 < shred00> all that i have tried anyway 14:18 -!- da-x [karrde@bzq-218-160-13.red.bezeqint.net] has quit [Read error: Connection reset by peer] 14:20 -!- Blaisorblade [~blaisorbl@host157-148.pool8289.interbusiness.it] has quit [Quit: Bye Bye - UML Rocks!] 14:23 -!- da-x [karrde@bzq-84-109-127-136.red.bezeqint.net] has joined #uml 15:18 < shred00> blech. Kernel panic - not syncing: Kernel mode fault at addr 0x14, ip 0xa008c845 16:31 -!- besonen [~besonen@dsl-db.pacinfo.com] has quit [Quit: Leaving] 17:07 -!- besonen [~besonen@dsl-db.pacinfo.com] has joined #uml 17:34 -!- ElectricElf [~david@electricelf.chair.oftc.net] has quit [Remote host closed the connection] 17:34 -!- ElectricElf [U2FsdGVkX1@electricelf.chair.oftc.net] has joined #uml 17:44 -!- Electric1lf [~david@149.9.0.8] has joined #uml 17:57 -!- ElectricElf [U2FsdGVkX1@electricelf.chair.oftc.net] has quit [Remote host closed the connection] 19:02 -!- loko [~rbrown@c-67-163-203-163.hsd1.pa.comcast.net] has joined #uml 19:50 -!- shred00 [~brian@linux.interlinx.bc.ca] has quit [Ping timeout: 480 seconds] 19:54 -!- lokkju [~lokkju@70-58-192-89.ptld.qwest.net] has joined #uml 20:07 -!- hfb [~hfb@adsl-69-231-52-189.dsl.irvnca.pacbell.net] has joined #uml 21:42 -!- Newsome [~sorenson@216-190-206-130.customer.csolutions.net] has quit [Quit: Leaving] 22:04 -!- hfb_ [~hfb@adsl-69-231-35-32.dsl.irvnca.pacbell.net] has joined #uml 22:11 -!- hfb [~hfb@adsl-69-231-52-189.dsl.irvnca.pacbell.net] has quit [Ping timeout: 480 seconds] 22:24 -!- Newsome [~sorenson@216-190-206-130.customer.csolutions.net] has joined #uml 23:00 -!- Nem^1 [~Nem@p54ABD5FE.dip.t-dialin.net] has joined #uml 23:07 -!- Nem^ [~Nem@p54ABF1C6.dip.t-dialin.net] has quit [Ping timeout: 480 seconds] 23:07 -!- Nem^ [~Nem@p54ABD5FE.dip.t-dialin.net] has joined #uml 23:07 -!- Nem^^ [~Nem@p54ABF1C6.dip.t-dialin.net] has quit [Ping timeout: 480 seconds] 23:07 -!- hfb_ [~hfb@adsl-69-231-35-32.dsl.irvnca.pacbell.net] has quit [Quit: Client exiting] 23:14 -!- Electric1lf is now known as ElectricElf 23:22 -!- Alecsandro [~ale@200.189.53.10] has joined #uml 23:27 -!- gorpon [~gorpon@adsl-69-226-1-158.dsl.pltn13.pacbell.net] has joined #uml 23:57 -!- timster [~chatzilla@64-142-81-224.dsl.static.sonic.net] has joined #uml --- Log closed Sun Jul 31 23:59:00 2005