up_the_irons: nor does bsd.rd -: up_the_irons files a ticket with toddf toddf: growfs isn't in the bsd.rd perhaps?
$ find /usr/src/distrib | xargs grep growfs
/usr/src/distrib/sets/lists/base/mi:./sbin/growfs
/usr/src/distrib/sets/lists/man/mi:./usr/share/man/cat8/growfs.0
but since you're already using growfs it means you have a fs to grow
implying you have a root fs around somewhere
mount /dev/wd0a /mnt
/mnt/sbin/growfs .. up_the_irons: toddf: yes, mounting an existing fs is what i ended up doing
I grew wd0a from 5G to 120G, now /bsd won't boot (hangs after "entry point at..."), yet bsd.mp boots fine
never seen this before. then again, i also never did growfs from 5G to 120G before. only did 40G max toddf: I bet the bios fails to read sectors after a specific size, I forget what the threshold is
I bet bsd.mp was around from before growfs, but you placed bsd after up_the_irons: toddf: both were around before growfs
toddf: i guess i should make / smaller then toddf: interesting up_the_irons: yeah i found it odd as well toddf: did you fsck after growfs? up_the_irons: toddf: yeah, went through clean
toddf: a clue: if I try to 'config -e /bsd' from a shell, it will never save it, instead says "not forced"
if I '-o bsd.new', i can get *that* one to save toddf: yeah, I normally do the dangerous way:
config -ef /bsd up_the_irons: then I 'mv bsd.new /bsd', but same symptoms
haha toddf: always have bsd.rd and/or obsd for rescue
but with your setup, there's the iso as well up_the_irons: yyup
i will re-image this fucker, and put the majority of space on /usr toddf: is this a template or one of your custom systems? up_the_irons: toddf: someone ordered a 120G OpenBSD VPS, which is rare (the 120G part), so I don't have templates for a 120G sized one. What I do is take the small one (5G), run the setup scripts, then _after_ it is all set up, go into disklabel / fdisk / growfs and change the disk size
toddf: this method has worked for 10G, 20G and 40G VMs, no problem. Never tried it on 120G toddf: heh, personally I have a pretty basic disk layout then leave the rest empty as a matter of customization and expansion later up_the_irons: yeah toddf: even the 10G disk has 7G free on my vps at your site up_the_irons: hehe toddf: OpenBSD area: 63-20964825; size: 10236.7M; free: 7734.4M up_the_irons: :)
you run a lean machine
toddf: do you like the new "Show" links for passwords in the Portal? ;) toddf: yes, I was gonna test them with more than midori, but yeah its much better than visible by default
next step, change-o-matic up_the_irons: right toddf: and for brownie points, permit a url for an iso for self install of vps's during order *grin* ***: Ujjain has joined #arpnetworks up_the_irons: i like much more for that kinda stuff to happen over a ssh pub-key protected interface (hey, maybe i'll put the change-o-matic in the console server)
toddf: one big problem with self-install or any type of auto-install / instant provisioning, is it attracts a lot of script kiddies; generally, if they have to wait 24 hours for something, or they get a sense "someone is watching", they go away ***: Ujjain has left toddf: I'd be happy for the functionality of changing things via ssh, so long as (via ssh or portal) one could submit an rsa key twiddle too ;-)
I honestly would like to be able to have multiple rsa keys in the .ssh/authorized_keys file for the console, I have a business, and I have an employee, and I do expect to be afk for !short periods of time in a few months
though I guess I could simply just make the rsa key be something both I and the employee have access to.. up_the_irons: toddf: it's almost like if i could spawn '$EDITOR ~/.ssh/authorized_keys' within the console menu, that'd solve that right there
so you can edit it yourself, even lock yourself out ;)
now if i could find an $EDITOR that doesn't allow shell commands to be executed... toddf: probably better to accept multiple lines of input for a replacement authorized_keys file, easier than finding an editor to suit your fancy up_the_irons: vi probably has a mode for this...
toddf: yeah
that's what i was thinking... that'd be pretty easy
the initial accounts are even set up that way. a script says "paste customer's key(s) below" ;) toddf: $ env SHELL=/usr/bin/false vi
date: exited with status 1
Press any key to continue [: to enter more ex commands]:
no guarantees this is a good thing to use though up_the_irons: echo "Paste user's public key below"
cat > $AUTHORIZED_KEYS_FILE
...
>> $AUTHORIZED_KEYS_FILE
toddf: i bet with a little bit of shell i could make it changeable, copy & paste fashion, no editing toddf: if the menu is a shell script up_the_irons: it is toddf: echo "pasta lavista:" up_the_irons: heh toddf: while read line; do [ "$line" = "EOF" ] && break; echo "$line" ; done > $AUTHORIZED_KEYS_FILE.tmp && mv $AUTHORIZED_KEYS_FILE.tmp $AUTHORIZED_KEYS_FILE
or somesuch
so e.g. if the ssh connection was interrupted during the paste it in theory wouldn't end up with a zero'ed file up_the_irons: right
there could also be "Now test your login, did it work? (y/N) : "
if that isn't answered with yes, then the old auth-key-file is restored toddf: sounds like a good safety net up_the_irons: yeah
re-imaged that VM, still wouldn't boot /bsd, bah
gonna keep / the same this time, and make a /usr toddf: maybe to keep the script kiddies away, you provision VM's with a 24hr min delay (or, set them up, delay the initial email) .. vs established customers who get a special merit badget of sorts that lets them provision more rapidly
aka I'm not exactly planning to provision more vm's, but if certain hardware or providers I have my physical systems at have issues, I'd be waiting with baited breath for the next vps to be setup
(fresh off the reality check, I had a dns server set to a client provided vm system, unfortunately I got very short notice at a bad time and am 4 days later just now pointing dns to your vps *grin*)
'short notice' -> "short notice the vm was going away"
something tells me that's a few steps down the automation path you might want eventually, I'm certainly not advocating you put it anywhere but the lowest rung of todo if it even has merit in your business plan ;-) up_the_irons: yes, all good points :) ***: schmir has joined #arpnetworks up_the_irons: toddf: do you new -O2 when you newfs your filesystems?
s/new/use/ toddf: -O2 is better for `bigger' filesystems, not sure where the threshold is
OpenBSD dos not yet support booting off of them
must be w/out newfs -O2
-O2 wastes a little space for lazy inode allocation
but makes faster fsck times and bigger fs's work fine
don't fall into the trap someone on the mailing list did recently
newfs -O2'ed a 6T partition and then couldn't fsck after lots of backup data was written ;-) up_the_irons: haha
good to know
tnx toddf: I have a 1T disk, sliced it into 250gb chunks just to be safe (though larger is fine, have a 500gb disk with 1 partition on a physical colo, it fscsks just fine regularly) up_the_irons: oh man, what is the openbsd equiv of this:
tar -cf - -C /mnt/old/ --numeric-owner . | tar -xpf - -C /mnt/new/ --numeric-owner toddf: there are other parameters to twiddle with larger fs's especially if the files are not small on average up_the_irons: bsd.rd version of 'tar' doesn't like the '-' for stdout it seems toddf: I'd imagine this should work on most anything:
(cd /mnt/old; tar cf - .) | (cd /mnt/new; tar xpf -)
if dump were on the boot media I'd say 'dump -0a -f - /mnt/old|(cd /mnt/new;restore -xf -' up_the_irons: ok, let me give that a try
it's doing something (didn't barf)...
nice
tnx toddf: I've used tar so much w/out -C I just fall back on old reliable habits w/out -C ;-) up_the_irons: toddf: i'd prefer to not use -C, it has always confused me; but to be honest, i've never fully groked tar itself, just so many little different things about it on every system toddf: I started with hpux, then slackware, debian, solaris, found openbsd and used that mostly, have some forced interaction with aix
basic stuff tends to work pretty reliably
not sure what --numeric-owner is about, I thought tar stored numbers by default instead of names up_the_irons: toddf: yeah but upon restore it likes to match up the right names accoring to /etc/passwd; a practice i don't like toddf: if it stores numbers how does it match up the names? up_the_irons: toddf: at least w/ GNU tar, probably not so on the *BSD versions. however the FreeBSD tar has that option
toddf: nfi, it may store the names too.. i just notice when i use --numeric-owner, i have a lot less problems upon restore ***: ziyourenxiang has joined #arpnetworks
bob^^ has quit IRC (Read error: Connection reset by peer)
bob^^ has joined #arpnetworks
baklava has quit IRC (Read error: Connection reset by peer)
baklava has joined #arpnetworks
mike-burns sets mode: +o up_the_irons
ziyourenxiang has quit IRC (Quit: ziyourenxiang)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
schmir has quit IRC (Remote host closed the connection) amdprophet: up_the_irons: are you running an ntpd within the arp network that is available? DaCa: he didn't the last time it was asked, just use pool.ntp.org amdprophet: kk
up_the_irons: why can i portsnap fetch extract so much faster from your network than my home network? i get like 200 KB/s down, but my connection is capable of so much more
do you have some ports caching server or something? ***: visinin has joined #arpnetworks up_the_irons: amdprophet: i don't have an ntp server
amdprophet: portsnap is faster from my network b/c it rocks
;)
srsly though, it is probably because of either (or both): 1) I'm close to some mirror, 2) My pipes are pretty fat (multiple GigE)
last time I checked, ftp4.freebsd.org was really close jeev: up_the_irons, dedicate a vps to a cvsup server
it's really easy
but i bet your under your commit so you dont care for the bandwidth use up_the_irons: yeah, i have a lot of unused capacity
but, nevertheless, it is "a good thing to do". link me to some docs on how to set it up ***: schmir has joined #arpnetworks toddf: http://www.openbsd.org/anoncvs.shar ? .. oh wait, thats openbsd sources ;-) up_the_irons: hehe toddf: amazing factoid about anoncvs + opencvs. the only files outside the CVSROOT are: 'touch dev/null' and a static binary of opencvs as 'usr/bin/cvs'
I can even serve anoncvs off afs in a chroot environment ;-) ***: Prael has joined #arpnetworks up_the_irons: that's pretty nice jeev: www.freebsddiary.org
it's really easy up_the_irons: roger jeev: http://www.freebsddiary.org/cvsupd-server.php
http://screamingelectron.org/forum/archive/index.php/t-292.html
cvsupd-server.php doesn't look like what i used
the second link is what i used but was on freebsdddiary
toddf, come on.. openbsd? we're not talking OS's that are in the same class with windows.. toddf: jeev: I have my fingers in my ears jeev: hahaha
you're gonna have to stick them into your eyes!
dood i've been working to fix someone's stupid xp virus for 4 hours now
i'm really good, if i could get into safe mode, i could remove the LSP. but cant
maybe i can find out where the lsp thingies are stored and delete from a PE ***: fastttputy has joined #arpnetworks fastttputy: hello, i am currently tring to optimizing all the stuff i can on my serve
server
to have the minimum delay for my users.
my server is on us, and trying to reponse to chinese poeple..
and ping are kinda very high..
what the best thing i could do to minimize delay and get faster reponse jeev: so you have a vps at arpnetworks and you want to minimize latency to your chinese users? fastttputy: oh wait.. this is only suppot for arpnetworks users?
=S
i thought its a support channel for networks ;( ***: cablehead has quit IRC (Ping timeout: 256 seconds)
visinin has quit IRC (Quit: asdf) jeev: networks ? heh fastttputy: well its more tcp ip optimizing
;P jeev: #networking or ##networking but good luck
latency is latency most of the time everything is ok i think fastttputy: hanks you!
hmm yaeh but i found some stuff tat help
as changing mtu
and putitng some dscp and tos on stack ip jeev: yea, you should try sites like dslreports.com and stuff like that, i dont think networking people will wanna help ;D fastttputy: except tat, i wonder if there more stuff to do :o lol
damn they suck >=(
so you mean they are selfish poepe :
:o jeev: no, think about it
people aren't getting paid to help...
and most of the time, it's repetative questions
you can't blame them for getting annoyed. ***: fastttputy has quit IRC (Ping timeout: 276 seconds)
cablehead has joined #arpnetworks
ChanServ sets mode: +o cablehead
schmir has quit IRC (Ping timeout: 256 seconds)
cablehead has quit IRC (Ping timeout: 248 seconds)
cablehead has joined #arpnetworks
ChanServ sets mode: +o cablehead
schmir has joined #arpnetworks up_the_irons: heh, networking support ***: darthn8ers has joined #arpnetworks
amdprophet|sleep has joined #arpnetworks
amdprophet has quit IRC (Read error: Connection reset by peer) darthn8ers: any admins alive in here? bob^^: what's up darthn8ers ? darthn8ers: I know some people who use your VPNs and referred me to ARP bob^^: i'm not an admin btw, just a customer :) darthn8ers: ahh :)
do you know if they support ubuntu 9.10 or debian 5 bob^^: i'm not sure i'm afraid :(
i run freebsd 8.0
i suspect you can probably upgrade a 9.04 system to 9.10 but i couldn't be sure darthn8ers: ok.. well thanks for your reply anyway
probably bob^^: debian 5 is supported though
it's on the order list
https://www.arpnetworks.com/order?product=vps&vps_special=1 darthn8ers: awesome.. I looked all over the site except the order page bob^^: hehe :) darthn8ers: I have never used a vps before.. is it kinda like using remote desktop? bob^^: it's basically a dedicated server
nothing installed except the operating system and ssh darthn8ers: so i wouldnt be able to use the gui in fedora? bob^^: well, if you ran it over vnc then yeah, probably darthn8ers: i really dont need the gui. I was just curious. bob^^: not sure how well it would perform though i guess darthn8ers: i am looking into a vps do download source code and compile with. then host the compiled .img for download.. would that be doable with a vps? bob^^: yup darthn8ers: sweet... thanks for your time bob bob^^: np :) amdprophet|sleep: darthn8ers: we're running ubuntu 9.10 ***: darthn8ers has left "Leaving" amdprophet|sleep: oops, too late..
lol bob^^: lol :) ***: amdprophet|sleep is now known as amdprophet amdprophet: weird... i don't remember setting my name to amdprophet|sleep jeev: your script probably did amdprophet: crazy linkinus -: amdprophet shakes fist ***: schmir has quit IRC (Remote host closed the connection)
ballen has joined #arpnetworks
ballen has quit IRC (Changing host)
ballen has joined #arpnetworks
ChanServ sets mode: +o ballen ballen: RandalSchwartz: let me know which one you end up liking the best. ***: ballen has quit IRC (Quit: Leaving...)
ballen has joined #arpnetworks
ballen has quit IRC (Changing host)
ballen has joined #arpnetworks
ChanServ sets mode: +o ballen
heavysixer_ has joined #arpnetworks
ChanServ sets mode: +o heavysixer_
heavysixer has quit IRC (Ping timeout: 260 seconds)
heavysixer_ is now known as heavysixer
heavysixer has quit IRC (Client Quit)
coil has quit IRC (Read error: Connection reset by peer)
coil has joined #arpnetworks