rhce-log-030625.txt

--- Log opened Wed Jun 25 16:54:08 2003 --- Log closed Wed Jun 25 16:54:46 2003 --- Log opened Wed Jun 25 19:02:55 2003 -!- sweeper [~mbroome@moya.trilug.org] has joined #trilug-rhce -!- Topic for #trilug-rhce: Next meeting: User and Host Security, Wed 25 Jun 2003, 8:00 PM EDT - IRC -!- Topic set by jtate [] [Wed Jun 25 17:13:09 2003] [Users #trilug-rhce] [@ChanServ] [ jeremyp] [ jtate] [ sweeper] [ Tribot] [ tripaste] -!- Irssi: #trilug-rhce: Total of 6 nicks [1 ops, 0 halfops, 0 voices, 5 normal] -!- Channel #trilug-rhce created Sun Apr 6 17:01:25 2003 -!- Irssi: Join to #trilug-rhce was synced in 0 secs < jtate> TallyHo sweeper -!- gmontag [~Eclectic@cpe-024-211-155-213.nc.rr.com] has joined #trilug-rhce -!- SinnerP [~sinner@ip68-101-8-79.nc.hr.cox.net] has joined #trilug-rhce < SinnerP> how-dee < jeremyp> Evening. -!- Nivex [mystic_nat@adsl-157-45-25.rdu.bellsouth.net] has joined #trilug-rhce < SinnerP> gutten evening -!- ovrclokd [~lisa@moya.trilug.org] has joined #trilug-rhce < ovrclokd> hello all < SinnerP> jell-o ovrclokd < ovrclokd> sorry i missed the meeting last week < ovrclokd> i made it back from tampa to charlotte, then got stuck in charlotte. :( < SinnerP> ovrclokd: a lot of airtravel sucks. The airports seem to be in a conspiration to cause delays, layovers and missing flights :( < jeremyp> My mother flies to Philly once or twice a week. The last three weeks she said she hasn't gotten home on time, or even within an hour of on-time, ONCE! < SinnerP> mmm < ovrclokd> sinnerp: absolutely. i was on 13 planes in 2 weeks recently, and about half of them had some issue or another < ovrclokd> jeremyp: my sympathies to your mother. OUCH. < SinnerP> maybe all those lay-offs of airport workers are somehow affecting the QoS < ovrclokd> sinnerp: i wouldn't doubt it. about half my problems were weather-based, but the other half were mechanical. < ovrclokd> sinnerp: it's amazing to me that i haven't fallen out of the sky. -!- cybertooth [~cybertoot@rdu57-31-133.nc.rr.com] has joined #trilug-rhce < jtate> cybertooth: You were kicking my butt in there. < jtate> So, does anyone have any questions about last week? -!- jtower [~jason@moya.trilug.org] has joined #trilug-rhce < cybertooth> Blame Jtower... he's my mentor < ovrclokd> i've seen /sbin/false used instead of /sbin/nologin. do they do the same thing, or does /sbin/false disable the account? < jtower> oops, forgot about this again, as usual < SinnerP> ovrclokd: my sister was on an airplane that had a propeller fell down, when taxiing < jtower> question - in top/uptime, what exactly does the load number actually correspond to? < ovrclokd> sinnerp: i think i'd have gotten off the plane at that point. :( -!- scott_l [~scott@user-0c8hpp4.cable.mindspring.com] has joined #trilug-rhce < jtower> a moose once bit my sister < SinnerP> answer: the number of processes waiting to be processed < jtower> ah < jtate> ovrclokd: not sure: their man pages are very different. < ovrclokd> sinnerp: it's the average for 1 minute, 5 minutes, and 10 minutes - right? < jtower> so it doesn't necessarily match up to cpu idle time < SinnerP> so, load < 1.0 , there's no waiting in processes to be executed < SinnerP> ovrclokd: sounds right. I don't remember the proper numbers < jtower> i have this on a client server: < jtower> 8:04pm up 40 days, 7:35, 1 user, load average: 1.35, 1.76, 1.33 < cybertooth> I've seen nologin used when they wanted to allow certain rights but no direct login to a box, false for when the user is simply used for a rights mask (like for an executing program), but they are similar enough to be interchangable < jtower> that seems pretty high to me < jtate> ovrclokd: nologin displays the message in /etc/nologin.txt and then exits. < SinnerP> jtower: this means that there are more processes tob e executed than the capabilities of your CPU < SinnerP> jtower: you want to check for any process going asstray and eating way too much CPU < ovrclokd> jtate: cool, so you can customize the "go away" banner. :) < jtate> false exits with an error code. < jtate> Right. < scott_l> is there a limit to the load ? ie: if a machine is just on it's knees is that load an absolute or relative number? could you see 9.9 as a load avg? < jtower> SinnerP: there are just a lot of different processes - apache, named, sendmail, ftp, ... < ovrclokd> jtate: will false still allow them to ftp / do something that doesn't need a shell login? < jtower> ovrclokd: yes < ovrclokd> scott_l: i've seen a load average of 53. (for real.) < scott_l> as well you can customize the "welcome" banner in /etc/issue I think is where that is done. < jtower> i usually change the login shell for people who only need email retrieval, for example < ovrclokd> scott_l: the box fell over shortly afterwards. < SinnerP> jtower: press 'P' to sort by CPU usage < scott_l> or remove it if just want a login in prompt and no info. < jtate> ovrclokd: what jtower said < SinnerP> and check who is eating CPU < scott_l> I was just curious. < SinnerP> I use /bin/false for email retrieving < jeremyp> scott_l: I've seen load averages in the hundreds before < jtate> SinnerP: Yeah, that's probably a good one. < jeremyp> On big FTP mirrors and such. Lots of FTP processes running and such < jeremyp> and such < jeremyp> :) < scott_l> ouch. < ovrclokd> jeremyp: d**n. < ovrclokd> jeremyp: was that a "normal" (as in, the box was running and serving users consistently, not on its way down) load? < jtower> one of my clients IMs me and says his web/mail server is very sluggish, and the loads are between 1 and 2. does that sound right? < jeremyp> Yes. kickstart.linux.ncsu.edu after a RHL release < SinnerP> jtower: nope < jeremyp> jtower: depends, run 'iostat' and see where the load is. If its in IDE drive access, then yes that could be sluggish < jtower> top shows that cpu idle time is hovering around 30-40% though < jeremyp> kickstart is all scsi, so its load was just due to the very high number of vsftpd processes running < SinnerP> jtower: which webmai program are they using? < jtower> bash: iostat: command not found < ovrclokd> jeremyp: makes sense. that's pretty impressive. < ovrclokd> jtower: /usr/bin/iostat < jtower> it's call an ensim server, it's a commercial virtual hosting package that runs on rh72 < jtower> butit uses standard daemons - apache, sendmail, proftp, and so on < cybertooth> I've played with a lot of ensim servers. The sendmail is setup poorly on the boxes < jtower> bash: /usr/bin/iostat: No such file or directory < jeremyp> ovrclokd: the limitation on the kickstart server was network bandwidth in or out < jtate> jtower: up2date iostat < jtower> jtate: no can do < jtate> Or however you install the package < jtower> it's not my server < jtower> although i do have root :) < ovrclokd> jtate: when you say set /tmp as its own filesystem, do you mean make it a separate partition? < jeremyp> ah, it's in the sysstat package < jtate> Hmm. ovrclokd what's the package name for that? < jtate> ovrclokd: Yes. < jeremyp> apt-get install sysstat < ovrclokd> jtate: is that just to keep someone who fills up /tmp from crashing the box, or are there other reasons? < jtower> this is not good: < jtate> That's one reason. < jtower> [root@host sbin]# df -h < jtower> Filesystem Size Used Avail Use% Mounted on < jtower> /dev/hda1 4.4G 4.4G 44M 100% / -!- RedWolfe [~ggw@durham-ar1-4-64-253-233.durham.dsl-verizon.net] has joined #trilug-rhce < jtate> jtower: Yes, that's a problem. < ovrclokd> jtower: not good any! < jeremyp> only 44M available ? < jtower> a big problem < scott_l> is that on the box that's slow? < jtower> well, i think that /dev/hda is only a virtual disk, not a physical one * RedWolfe wave meekly < ovrclokd> hi redwolfe! your brother is on his way home. :) < jeremyp> hda = IDE ... IDE raid? or something strange with ensim ? < jtower> the latter < jtate> http://www.trilug.org/~lisa/rhce-security-talk.030618.txt are the notes from last week. < RedWolfe> merci beaucoupe < jtower> no way a P4 server is running a 4gb ide disk < SinnerP> jeremyp: RAID is /dev/mdx < jeremyp> SinnerP: I meant hardware IDE raid < jtower> i think ensim "tricks" each virtual server into thinking it has /dev/hda of its own < jeremyp> jtower: that just means the first partition is 4GB, not the whole disk < jeremyp> jtower: run 'fdisk -l /dev/hda' < SinnerP> then the device depends on the driver (module) < jtower> that's the ONLY partition < jtower> fdisk and parted don't work < jeremyp> (that's safe, it just listens the partition table) < jtower> that's why i think its a virtual disk < jeremyp> no swap shows? < jtower> no < jtower> still, its a FULL virtual disk < jtower> it's funky < jeremyp> does 'free' show a swap partition ? < SinnerP> jtower: for some reason, I believe that /dev/hda cannot be a virtual disk < jtower> [root@host sbin]# free < jtower> total used free shared buffers cached < jtower> Mem: 262144 112228 149916 28696 0 18292 < jtower> -/+ buffers/cache: 93936 168208 < jtower> Swap: 131072 25828 105244 < jeremyp> SinnerP: it is, in the context of ensim < jtate> jtower: Look for a link in /dev < SinnerP> jeremyp: wierd < jeremyp> Looks like ensim uses kernel patches to do wacky stuff. I've heard of this a little over on shrike-list < SinnerP> jtower: cat /proc/memstat < jtower> [root@host dev]# uname -r < jtower> 2.2.19-6.2.1ensim-3.1.5-2 < ovrclokd> sinnerp: d'you mean /proc/meminfo? < jtower> jeremy shoots and scores < jtower> 2.2 kernel no less < SinnerP> ovrclokd: cat /proc/mem[TAB] < ovrclokd> sinnerp: *grin* < ovrclokd> sinnerp: just wondering if there's something that should be in my /proc that isn't! < SinnerP> hehehe < jtower> jtate: there is a /dev/hda, not a symlink, but a block device < jtate> Hmm. < jtower> i mean, i can tell what the problem is. the disk is full and the server is being hammered < ovrclokd> jtower: what's in /proc/partitions? < jtower> that's all there is to it. 120+ processes, mostly httpd and sendmail < ovrclokd> jtower: what's filling it up? can you relo some logs? < jtower> proc is mostly empty, none of the usual stuff is there < jtate> du -h might be useful < jtower> ovrclokd: hard to tell, du won't return any results < SinnerP> ovrclokd: a list of your partitions < jtate> Doh. Nevermind. < jtower> and i haven't poked around fully yet < SinnerP> ovrclokd: /proc is very very useful to get *real* info about the systems < ovrclokd> jtower: you could check /etc/httpd/logs < ovrclokd> sinnerp: yes - i was wondering what it would tell jtower about the partitioning on the slow box < jtower> i'm trying to talk the guy into letting me build him a new server, one that can handle the load better < jtate> jtower: Where are the mail spools? < jtower> no idea < jtower> ensim does some wierd things with users < jtower> and their mail and other personal files < RedWolfe> ? ensim ? < cybertooth> The spools are in the chroot. < jtower> it's not meant to be managed via a shell, but rather by the web interface < jtower> it's like a twilight zone linux server < SinnerP> jeeeezzzzzz -!- ccw [~ccw@durham-ar1-4-64-253-233.durham.dsl-verizon.net] has joined #trilug-rhce < ovrclokd> jtower: try home/virtual/$siteadmin/var/spool/mail/$user < ovrclokd> oops, dropped the leading /, sorry < jtower> yup, that's where they are < jtower> bunch of symlinks in /home/virtual * SinnerP has a list of LDAP tools * jtate wants a them < ovrclokd> ensim docs says there's a cron job to move the mail to /var/spool/mail/$user every 5 min, too < jtower> SinnerP: where were you and your list last night when we needed it? < cybertooth> All the ensim servers I've worked on were risc processors ... very slow. < cybertooth> We did okay! < SinnerP> jtower: ... speeping? < cybertooth> better than sleeling < jtower> SinnerP: just joking. we had a minor ldap problem last night on a suse box < jtower> we == cybertooth and i * jtate wishes he could have gone < jtower> jtate: it was pretty slick, although the installer was a tad buggy < SinnerP> gq <http://biot.com/gq/> , Directory Administrator <http://diradmin.open-it.org/index.php> , KDirAdmin <http://www.carillonis.com/kdiradm/> , LDAP Browser/Editor, <http://www-unix.mcs.anl.gov/~gawor/ldap/download.html> , LDAPExplorerTool, <http://ldaptool.sourceforge.net/> < ovrclokd> jtower: there's an ensim webappliance pdf guide at http://www.remarkablehosting.com/support/guides/wp23/WEBpplianceTechGuideBook.pdf < cybertooth> josep, how about sending it to the rhce list? < jtower> ovrclokd: neat, unfortunately that's for a really old version < ovrclokd> jtower: d'oh! sorry. < SinnerP> cybertooth: good idea < SinnerP> (doh!) < ovrclokd> jtower: what version are you running? < SinnerP> cybertooth: who are you again? < jtate> Of course it's a good idea < cybertooth> tribot, who am i? < Tribot> cybertooth is the amazing and incredible Jon Carnes < RedWolfe> tribot: whois cybertooth < Tribot> cybertooth is the amazing and incredible Jon Carnes < cybertooth> jinx! < SinnerP> doh! < SinnerP> XDDDDDDDD < jtate> Tribot: Who is amazing and incredible? < Tribot> jtate: no idea < jtower> ovrclokd: 3 point something < jtower> and it's nearly two years old < jeremyp> how many virtual servers? < jtate> So, I think that we decided that a misconfigured /etc/hosts.allow & /etc/hosts.deny would be a great thing for the troubleshooting part of the RHCE < jeremyp> yeah, misconfigured network files in general.. .hosts, exports, network-scripts/ifcfg-eth* < jtate> Note that there are a ton of ways to screw up access to your system: < ovrclokd> jtower: if you google for ensim webppliance technical guide 3 there're several versions... < jeremyp> oh, and make sure you know how to do password recovery < jeremyp> does everyone know how to do that? (password recovery) < jeremyp> ie, boot in single user mode < jtower> ovrclokd: it's not actually my server to admin, some other company runs it for my client. i just have the root pw for his virtual server, not the entire box < jtate> tcpwrappers, xinetd, iptables as well as the individual application can limit access based on ip address. < scott_l> I was thinking that was the answer, (single boot) I was sitting here wondering if there was another way < jtate> scott_l: a reinstall, if you have /home in a separate partition. < jeremyp> use 'a' in grub to add 'single' to the end of the boot line < jtate> Tom's root boot. < ovrclokd> jtower: mm, so you're supposed to debug it without having full access to it. i hope you're charging him enough. :) < jtate> If you < jeremyp> or "linux single" in LILO (assuming the boot image is called linux) < jtate> 've forgotten the grub password. < jtate> The Rescue disk should give you enough to get in to recover your password. < jeremyp> Yes but you might not have the rescue disk or Tom's Root Boot on the RHCE < ovrclokd> jeremyp: will they let you take tom's root boot in with you? < jeremyp> Kknow single user mode! < jtower> ovrclokd: not really, he just asked me to look into it < ovrclokd> :) < jtower> but i did tell him that his disk "slice" was totally full < ovrclokd> jtower: that's fair. i'm just punchy tonight - had a cider with dinner. < RedWolfe> no take-in's on RHCE, but the boot images might be available < SinnerP> jeremyp: I thought that you did "e" (edit) and then added " single" to the end of grub's boot line < jtower> ovrclokd: besides, i received his last big invoice payment today. that'll buy him some free troubleshooting :) * SinnerP has just sent the email < jeremyp> SinnerP: with e, you have to choose a line to edit and it's a pain < jeremyp> SinnerP: with a, it's much easier < ovrclokd> jtower: always a good thing. are you payday splurging at stayonline tomorrow? :) < SinnerP> jeremyp: just edit the longest line. It always works (TM) < jtower> ovrclokd: sure, meet me there at noon. if i'm a litle bit late just keep waiting... < ccw> I found a fairly obsure way of lousing up X this week (talking about obsure ways of boluxing a system)... < jeremyp> SinnerP: 'a' is much easier. Why make it difficult ? < SinnerP> jeremyp: I knew 'e'. I didn't knew 'a' < jtower> isn't e for making permanent changes, while a is only one time? < SinnerP> 'e' is for editing the whole line... just that time < SinnerP> so you can change boot partitions and kernel versions < SinnerP> useful when grub.conf is botched < jeremyp> I suppose you could use e if you really messed up your grub.conf < ovrclokd> jtower: already indulged in my splurge for this month - 2 new WD1200JBs. :) < jtate> ovrclokd: Are they the same Firmware? < ovrclokd> ccw: whatja do? < SinnerP> woohoo! < RedWolfe> ccw: so tell us already < jeremyp> but a is useful for adding/editing the kernel parameters, namely adding 'single' (or other test kernel parameters), so it's faster and easier for most purposes < jtower> ovrclokd: i have one of those in my p166 backup server < jtate> I've got two of those, but different firmware, and one drive has 31MB more than the other. < jeremyp> ovrclokd: 120GB? Wussy! Go for the 160 ! :_ < ccw> the directory /tmp wasn't umask 1777 (it was 755). X wouldn't start because xfs couldn't make the unix/7100 socket and X couldn't find the "fixed" font and wouldn't start ... < jeremyp> (that was supposed to be a :) < jtower> jtate: want a matching set? i'll see what firmware mine has and trade you if you want < SinnerP> ccw: that's a good trick < jtate> Nah, I' < ovrclokd> jtate: dunno - i'll have to check! how do you tell what firmware it has? < jtate> ve already got the mirror all set up. < SinnerP> another good trick is to have another :0 X session runing... in VT != 7 , < jtate> ovrclokd: You'll have to boot using the WD diag disk. < ccw> it seems /tmp got set to 755 during RedHat 8.x to 9.0 upgrade ... < ccw> It took me a little while to figure that one out ... :) < ovrclokd> jeremyp: just couldn't bring myself to pony up the $$$. < RedWolfe> oh, RH claims they dont support /usr/tmp as anything but a symlink the /var/tmp! < jeremyp> ccw: that's odd, it should have been kept with the sticky bit < ccw> jeremyp: sounds like famous last words... :) it was 1777 when we started and 755 when we finished ... < jtate> There are some nifty LDAP scripts in /usr/share/openldap/migration < jtate> ccw: Going back in time are we? * ccw wonders how so ... * RedWolfe does the "Time Warp" < ovrclokd> jeremyp: the 120Gb were the same price i paid for my pair of 80s last year. ah, progress... * ccw only does Rocky Horror on Fridays at the Realto ... < jtate> Remember though you still have to set up the root entry in LDAP before you can add any of the output generated by those scripts. < SinnerP> jtate: yes, there are. And MDK has an interesting tutorial for LDAP + Samba integration, by "Ranger" < SinnerP> jtate: very true < SinnerP> ccw: you going to the Realto? < ovrclokd> jtate: is that their data lifeguard suite? i bought mine bare, so no disk - have to see if i can download it. < jtate> ovrclokd: Yeah, that's the one < jtate> ovrclokd: Sadly, there's no utility to update the firmware. < cybertooth> think I'm just going to bz for a bit... < ovrclokd> cybertooth: have fun storming the castle! :) < ovrclokd> jtate: bummer. :( < RedWolfe> s/Realto/Rialto/ -!- scott_l [~scott@user-0c8hpp4.cable.mindspring.com] has left #trilug-rhce [] < ccw> SinnerP: sorry, went and got a Coke. Rocky Horror at the RIALTO every Friday night at midnight ... < SinnerP> cool < SinnerP> say hi to mikem * SinnerP goes for an iced tea * jtate had an ex-girlfriend who used to go pretty frequently. < ccw> SinnerP: don't forget your newspapers, umbrellas, rice, bics, what have you ... but I haven't been down there in a while ... < jeremyp> jtate: um, not sure we wanna know that :) < SinnerP> jejeje * ccw too much information ... < SinnerP> ...data overload < jtate> Go to Rocky * RedWolfe does the "Time Warp" < ccw> we all could go Friday week ... ? * ccw imagines an RHCE Rocky Horror excurssion ... < SinnerP> my Fridays are pretty busy: drivin' drivin' and more drivin' < ccw> At midnight ?? < SinnerP> RHCE Rocky Horror Certified Engineer ? < RedWolfe> Sinner, going somewhere and back? < SinnerP> RedWolfe: Winston-Salem <-> Greenville < SinnerP> ccw: sometimes, I finish driving at imdnight, yes < RedWolfe> Rocky Horror Certified Emoters < SinnerP> sometimes, Greenville <-> Richmond < ccw> EVERY Friday ?? < SinnerP> ccw: most of them < SinnerP> I got awesome mileage, though < RedWolfe> ovrclokd, any special questions? < ovrclokd> sinnerp: sounds like you need a hybrid. < SinnerP> ovrclokd: better than that, I have a TDi < ccw> Well, does anybody else want to join in a RHCE RHE? < RedWolfe> ?TDi < ovrclokd> redwolfe: i looked over the notes; they look pretty thorough. < SinnerP> RedWolfe: Audi's Turbo Direct Injection diesel engine < RedWolfe> Ahh, cool < SinnerP> RedWolfe: on a VW Jetta. < ovrclokd> redwolfe: is there any way other than sudo to give a non-root user the ability to mount a drive? < jtate> Automount in /etc/fstab < RedWolfe> yep, usermount option in fstab < jtate> You could probably also figure out a way to put the drive in /etc/security/console.perms < ccw> write a SUID program that mounts the drive, unSUIDs and spawns a shell. Exit to unmount (scripts have too much potential for abuse...). * ccw like jtates solution, too... < ovrclokd> cool ideas. what about smbmounting a share on from another system? < ccw> ovrclokd: me or jtate? < SinnerP> you can specify that on /etc/fstab . same rules < ovrclokd> ccw: anyone. :) < jtate> ovrclokd: LUFS is supposed to allow user space mounting. I thought there was an option for using smbmount as non-root. < ccw> ovrcclokd: want a generalizedd solution or fs specific? < ovrclokd> ccw: yes. :) < ovrclokd> ccw: i smbmount a share on mike's box to backup my digital images over thre < ovrclokd> ccw: it'd prefer not to have to su to root every time to mount/unmount it < ovrclokd> s/it'd/i'd < RedWolfe> ovrclokd, the user option should work < jtate> Of course you have to be root to edit the /etc/fstab. < jtate> You could setup smbmount to be a console app. < RedWolfe> that wasn't the problem, she didn't want to have to keep using su/sudo < jtate> check out the files in /etc/security/console.apps < ccw> ovrclokd: in that situation, automount ought to we just what you'd need... < jtate> These are the applications that a "Console" user can run. < jtate> A console user is the first person to log into a machine either on the TTY or GDM/KDM login screen. < ovrclokd> jtate: hm. does that include all shell sessions for the logged-in user? < ovrclokd> ccw: is there any way to force the drive to unmount when i'm done with it? < jtate> ovrclokd: It should. < ovrclokd> ccw: i know automount would unmount it eventually, but i don't want it hanging out there in case mike's box goes away -!- Nivex [mystic_nat@adsl-157-45-25.rdu.bellsouth.net] has left #trilug-rhce ["Exit, stage right."] < ccw> you can set the timeout period in fstab ... < jtate> It sets perms on the files, no real magic. < ovrclokd> jtate: sweet. i'll give that a try. < SinnerP> jtate: Console user... is it always the same user? or is it the first user that logs in after a reboot? < SinnerP> can root be the Console user? < ovrclokd> ccw: mm, yes, but if i wander off while it's transferring, it might unmount before i'm ready < jtate> The first one to log in, until all sessions for that user are terminated. < jtate> Yes root can be the console user. < ovrclokd> ccw: altho i guess if i try to access the directory it'll just re-mount it for me... < jeremyp> if another user subsequently logs in, they'll get the perms < jeremyp> rebooting doesn't have anything to do with it < ccw> that's whoever logged in on console until session token is relinquished, I think ... < RedWolfe> automount monitors activity and backs off if the fs is busy < RedWolfe> ccw, right < ovrclokd> redwolfe: *nods* i tend to kick off a copy, walk away, and check on it in 15 minutes or so. sometimes it only takes 3 of those minutes, tho. :) < RedWolfe> it is limited to the physical console * ccw nods to RW comment ... < ovrclokd> jtate: i have redhat-config-services in /etc/console.apps - when i try to run it as lisa, it prompts me for root's pwd. is that because user=root is set in the file itself? < ccw> OC: will not unmount if fs is in use. This includes current directories of a shell ... so, if you cd to fs, it will < ccw> stay mounted until you cd out or exit ... < jtate> ovrclokd: Yeah. Probably. < ovrclokd> jtate: hm. just changed to user=lisa and i'm still getting the prompt for root's password. interesting... < ccw> OC: you may have to restart the service ... < jtate> Comment out the user= line, then log out of all consoles and log back in. < ccw> or do what JT said ... < RedWolfe> the rh-config-* programs use "userhelper" to get admin permissions, if you set sbin /usr/sbin before bin /usr/bin you'll get the real command * ccw mumbles for a moment and then says, "or not"... < ovrclokd> jtate: i'll try that after we're done, then - this is running in one of the consoles! :) (don't want to lose my log.) < ccw> BTW, is anybody a GRUB expert? I got'a problem ... -!- jimstigator [~jim@moya.trilug.org] has joined #trilug-rhce < jtate> ccw: What's the problem? < ovrclokd> thanks for your help, folks... :) < SinnerP> I'm going to leave now < ccw> GRUB won't automatically boot anymore. We bought a new HD and it came with a 133 Mbit IDE controller... < jtate> I'm not an expert, but hopefully I can help. < SinnerP> I gotta get up waay too early < SinnerP> see you all next time -!- SinnerP [~sinner@ip68-101-8-79.nc.hr.cox.net] has quit ["g'night"] < ovrclokd> g'night sinner! < RedWolfe> Sinner: happy driving < jtate> ccw: You've probably got a HD controller order problem with your BIOS. < ccw> We installed it, and put the IDEs on it and the CD-ROM and CD-RW on the old MB IDEs. Now, it won't boot without intervention ... < RedWolfe> is there a way to make the off-board controller the first? (prob depends on the bios :( < jtate> ccw: intervention meaning boot floppy or CD? < ccw> jtate: OK, how do I fix it??? < RedWolfe> hand typing at GRUB or boot floppy < jtate> RedWolfe: Yes, it'll depend on the bios. < jtate> RedWolfe: So wait, the GRUB prompt comes up? < ccw> Meaning that it comes up with a GRUB promt and I have to type "configfile /grub.conf" <return> < jtate> Hmm. Is your old drive still /dev/hda or whatever? < RedWolfe> hda=cd-rom hdc=cdrw hde=old drive hdg=new drive < RedWolfe> no b,d,f < jtate> Ok, in /etc/grub.conf what's the last commented out line before the start of the config? In mine, it's /dev/hda < ccw> not, old drive was 100Mbit so we put it on the new controller... < ccw> wait one ... < ccw> #boot=/dev/hde < RedWolfe> says #/dev/hde ids as (hd0,0) fine < jtate> What was the old drive's letter before the HW upgrade? < ccw> /dev/hda < ccw> have re-insalled GRUB ... < RedWolfe> (several times) < ccw> (two, three times in fact) < jtate> What did you use to do that? < ccw> Grub < RedWolfe> grub-install and grub < jtate> so you did grub-install /dev/hde < RedWolfe> (grub-install being a script frontend for grub) yes < ccw> basically, said "info grub" and followed the directions for "natively installing" ... < jtate> And you're sure that (hd0,0) is correct? < RedWolfe> probs constatnly say so < RedWolfe> s/constatnly/consistently/ < ccw> info grub->installation->installing GRUB natively < RedWolfe> s/probs/probes/ < ccw> configfile /grub.conf finds it fine ... < jtate> Hmm. < ccw> kernel and root commands work peachy kean coool ... < ccw> with (hd0,0) < jtate> Do you have a /boot partition? < ccw> yes, in an MD0 < ccw> RAID1 < ccw> set "e2label" /boot < jtate> Did you have this before the HW upgrade? Or is this new? < ccw> machine on the desk to the right of me has same but (HD0,0) is /dev/hda ... < ccw> that is MD0 is /dev/hda and /dev/hdb ... < jtate> I couldn't get that kind of setup to work under Grub. Had to use LILO. * ccw looks at tembo machine and tells it that it shouldn't be working. It ignores him ... * RedWolfe grins < tripaste> "jtate" at 127.0.0.1 pasted "My grub.conf" at http://members.trilug.org/tripaste/5 < jtate> It seems to me that grub is installed correctly. Maybe you have the wrong path somewhere in your conf file? * ccw thinks jtate probably is right and goes to look for a bigger hammer ... < jtate> Can you post your grub.conf file to http://members.trilug.org/tripaste ? < RedWolfe> RW thinks our confs are equiv except for hde < ccw> hang on 2 seconds... < ccw> Is now in http://wolves.homeip.net/~ccw/grub/grub.conf < ccw> note, comment ain't the exact truth. Anaconda was run before IDE was changed ... I think (am I right RedWolfe?). < RedWolfe> yes, hw change came after upgrade to RHL9 * ccw wonders idlely if anaconda can be run without running installation... < jtate> So your boot partition is the first partition on /dev/hde * ccw corrects himself. CCW wonders if the Disk Configuration GUI can be run seperately ... < jtate> Nope. < ccw> Jtate:correct < jtate> Yeah, that's as far as I got with it. I switched to lilo and it works fine now. < jtate> Sorry I can't be more help. Perhaps Jeremyp can be of more assistance. < jtate> I've got to go. -!- jtate [~jtate@rdu74-181-041.nc.rr.com] has left #trilug-rhce ["Laters"] < RedWolfe> well, RW and so want to go karaoke.... < ccw> fdisk -l output in http://wolves.homeip.net/~ccw/grub/fdisk-l.hde.txt and http://wolves/homeip.net/~ccw/grub/fdisk-l.hdg.txt < RedWolfe> ping jeremyp !! < ovrclokd> yeah, i've gotta jet, too. hope you guys manage to wrestle grub into submission some day. < ccw> oops, fdisk -l output in http://wolves.homeip.net/~ccw/grub/fdisk-l.hde.txt and http://wolves.homeip.net/~ccw/grub/fdisk-l.hdg.txt -!- ovrclokd [~lisa@moya.trilug.org] has left #trilug-rhce [] < jeremyp> RedWolfe: pong, sorry * ccw wonders if the RHCE logging is over ... * ccw wonders if anybody is there ... * ccw sings: Does anybody know what time it is ... ? Does anybody really care ... ? * sweeper is still logging, but hasn't been paying any attention < sweeper> jeremyp: is that you typing? it doesn't show a username * ccw waves a hand in front of jate's glassy stare ... < ccw> look to the immediate right of the asterisk for the user. i.e. me. * sweeper just now notices that his color scheme makes some of the irssi highlighting unreadable < sweeper> I see you now :) < ccw> effect is produced by typing /me the rest of the text ... < sweeper> anywho ... seems like things are pretty much done so I'm logging off. Ciao! * ccw waves to sweeper as he departs ... * ccw thinks he has gotten as much help with his GRUB problem as he's going to get ... --- Log closed Wed Jun 25 22:36:51 2003 Generated by irclog2html.pl 2.1 by Jeff Waugh - find it at freshmeat.net!