Re: Moving LVM volume?
On Jan 02 11:52, Joel Rees (joel.r...@gmail.com) wrote: There can be reasons to do such things on production servers, but even then, you are basically only going to do it to salvage data, never the OS. LVM is also a great way to avoid downtime when adding or replacing disks. A couple months ago I had to replace all disks (24 of them, in three RAID controllers) in a file server, and with LVM it could be done with zero downtime (OK, I did schedule a reboot after it was done just to make sure it worked). (I've been a great fan of LVM ever since 1992 or thereabouts, when it was first introduced in HP-UX. It's a major boon in systems with multiple disks and long lifetimes, less so in workstations - although I find it useful enough even in them, as disks tend to need replacing if only because they get too small, and it's nice to be able to do that without much downtime.) -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20150102084211.ga25...@tarvainen.info
Re: Any advice for a user about to use LVM for the first time?
On Jan 01 23:42, Bob Proulx (b...@proulx.com) wrote: Using mdadm RAID? Or LVM raid? I have personally only used mdadm raid and not lvm raid. Yes. LVM raid is today usually best to avoid. I generally use mdadm for raid (unless hardware raid is available) and put lvm on top of that. And I tend to use raid whenever I can, that is, in almost every machine with enough room for disks, and raid1, raid6 or raid10, sometimes raid0 when speed is crucial but data is expendable (cheap to regenerate); raid5 I try to avoid. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20150102201305.ga22...@tarvainen.info
Re: Moving LVM volume?
On Jan 01 01:54, Frank Miles (f...@u.washington.edu) wrote:
I recently added a new hard drive to my home system. I decided to use it
to create an all-new bootable 'jessie' system. I created a partition
table that I thought would be flexible:
/dev/sdb1 / (root) {7G}
/dev/sdb2 /swap {4GB}
/dev/sdb3 /oldjunk{1G}
/dev/sdb4 extended {remainder}
/dev/sdb5 LVM{one large volume}
Most of the partitions- /usr, /home, /var, ... were in LVM2.
What I've learned since then is that /usr seems to have special
status, and probably shouldn't be part of LVM as certain tasks
early in the boot process can't seem to access the interior of
LVM.
That's strange. I've been having /usr as well as / and swap, indeed
everything except /boot on LVM in countless machines since I've
forgotten when, without any particular problems (and I've often
been happy I could enlarge / or /usr without even booting).
Perhaps you could explain what are the problems you see or
anticipate with /usr under LVM?
But if you insist, a few options:
(1) You have a pretty big root partition. Perhaps you could simply
enlarge it a bit and make do without a separate /usr?
Just move swap and oldjunk to LVM, enlarge root to 12GB,
move /usr there, done.
I've moved 'oldjunk' into the LVM, and want to expand this
partition to become the new /usr. I've shrunk the LVM, but
the freed space is all at the far end of the LVM. I have
been unable to move it towards the end of the disk space,
so I can expand /dev/sdb3. gparted, resize2fs, pvmove,...
(running from a CDROM-based rescue disk) have all failed.
(2) After shrinking LVM, create a new partition after it
and put /usr there.
(3) Take a fresh backup, reinstall with desired partitions,
restore. (You *do* have proper backups, right?)
(4) If you *really* want to move the LVM partition, it can
done but it's a seriously messy and error-prone operation.
None of the standard tools will do it automatically, AFAIK.
Outline: make sure you have a fresh backup, boot off rescue disk, make
sure LVM isn't active, move the contents inside the LVM partition in
as small pieces as necessary (depending on how much free space there
is) using dd or similar, taking care of aligments c, then delete and
recreate partitions to match new locations (taking care not to let any
too automatic tool to mess things up), reboot, realize you've
messed up, go to step (3). :-)
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150101101305.ga2...@tarvainen.info
Re: Bash Script for Mail sending Check
On Sat, Apr 18, 2009 at 08:27:43AM +0300, Andrei Popescu (andreimpope...@gmail.com) wrote: On Sat,18.Apr.09, 08:04:12, Kurian Thayil wrote: only whether the port is open or not. But what I need is to check if an SMTP server is listening in that port. So I guess the best tool is telnet and to check 220 status. Hhm, I thought a port cannot be open unless some server is *listening* on that port. What other server do you think would otherwise listen to port 25? It could be under inetd, which would keep the port open even if it couldn't actually start the mail server, or there could be a mail server that's not accepting mail for some reason. And people do abuse ports - I've even seen ssh in port 25. So if the intent is to check if mail could be sent, checking for smtp reply code is better. (It's not 100% reliable either, but then nothing is, short of actually sending the mail and waiting for reply or something.) -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Bash Script for Mail sending Check
On Thu, Apr 16, 2009 at 11:28:10AM +0530, Kurian Thayil
(kurianmtha...@gmail.com) wrote:
I got your modification working only when telnet gives a prompt,
Right you are - telnet doesn't react to the escape sequence
before connecting. Bummer. :-(
Oh well, just kill it then:
#! /bin/bash
case $( (sleep 5; echo quit; ps |awk '/telnet/{print kill $1}' |sh) | telnet
192.168.0.20 25 2- )
in
*'
220 '*) echo Success ;;
*)echo Failed ;;
esac
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Bash Script for Mail sending Check
On Thu, Apr 16, 2009 at 07:56:04PM +1000, Alex Samad (a...@samad.com.au) wrote: can you use nmap ? It will tell you if the port is open or not, but not whether there is a mail server behind it willing to talk to you. I guess you could first use nmap to check the port and only then try telnetting to it. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Bash Script for Mail sending Check
On Thu, Apr 16, 2009 at 03:38:23PM +0200, Brent Clark
(brentgclarkl...@gmail.com) wrote:
what Im really trying to say is that your
code may look a lot neater and easier, as opposed to
case $( (sleep 5; echo quit; ps |awk '/telnet/{print kill $1}' |sh) |
telnet 192.168.0.20 25 2- )
nc -v -w 5 -z 192.168.0.20 25
Neater, yes, but the functionality is not equivalent.
As I understood it, OP specifically wanted to test if they
can connect to the mail server, not just if port 25 is open.
Also, common though nc nowadays is, it's still not present
in every system, and if you have to install a new program
for the purpose you might as well install swaks.
Finally, irrelevant though it may be to Debian, portability
will suffer from every new external program you need.
Ugly though that kill hack is, it'll work in more or less
every *ix-related box, as it only uses standard POSIX tools.
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Bash Script for Mail sending Check
On Wed, Apr 15, 2009 at 11:47:43PM +0530, Kurian Thayil (kurianmtha...@gmail.com) wrote: [...] I need to execute a script only after making sure that I am able to send mail using the relay server 192.168.0.20 (Postfix enabled). Is it possible for me to write a bash script which check this? Doing it in a general, reliable way with bash alone would be very difficult, but if it's a local server that always responds fast or not at all, you could try something like this: #! /bin/bash case $( (sleep 5; echo quit) | telnet 192.168.0.20 25 2- ) in *' 220 '*) echo Success ;; *)echo Failed ;; esac -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: how can i turn /dev/null into an MTA?
On Wed, Apr 15, 2009 at 07:34:29PM +, Tzafrir Cohen (tzaf...@cohens.org.il) wrote: BTW: is it possible to configure exim (or sendmail) not to run as daemons? Don't know about exim but for sendmail it is perfectly possible, well-supported and indeed easy. Whether that is a useful configuration depends on your needs, but I've used it with success in situations where client programs can execute sendmail binary to send mail as an alternative to via port 25 - surprisingly many have that option. In Debian the right way to do this is editing /etc/mail/sendmail.conf (set DAEMON_MODE=none). -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Bash Script for Mail sending Check
On Thu, Apr 16, 2009 at 09:58:37AM +0530, Kurian Thayil (kurianmtha...@gmail.com) wrote: I tried the script. Good work. But, one problem though. If the mail server is firewalled (Netfilter) and if its blocking the client (one where bash script is run - 192.168.0.10), then `telnet 192.168.0.20 25` will not give any reply but just tries to connect indefinitely. At that moment the script will not give the error message. Can we do anything about it? You could abort the telnet using its escape sequence instead of using smtp quit, i.e., #! /bin/bash case $( (sleep 5; echo -e '\033q') | telnet 192.168.0.20 25 2- ) in *' 220 '*) echo Success ;; *)echo Failed ;; esac If you want to be friendly to the mail server you could do both: case $( (sleep 5; echo quit; sleep 5; echo -e '\033q') | telnet 192.168.0.20 25 2- ) in That will look a bit nicer in the mail server logs and may avoid tripping some spam/attack detection tools. Also, you may need to adjust the delays: the time a mail server takes to respond to initial connection may vary a lot, depending on things like its load and whatnot. Quite a few even deliberately wait some 10 seconds just to trip spamming software known to be too impatient. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
GlusterFS (was Re: WinBlow$ Home Server equivalent)
On Sun, Apr 12, 2009 at 04:08:30PM -0400, Stefan Monnier (monn...@iro.umontreal.ca) wrote: If someone has actual experience about dealing with lost disk in WHS, please tell us. Actually, I'd rather you don't tell us: just tell him. This is a mailing list dedicated to Debian, so Windows discussions are off-topic. Are comparisons with other OSes also considered off-topic here? If so, my apologies. That's anyway my only interest here, I'm just curious when I heard claims Windows can do something Linux can't. But, more on-topic: it just occurred to me the suggested (hypothetical) functionality could be achieved with GlusterFS. I haven't actually tried it, but it should allow for a setup where each disk has its own filesystem from the server's point of view yet are shown as a single whole to the clients. Every individual file would be on only one disk, but there'd be no need to decide manually which goes where. If you pull a disk out, only files on that disk are lost - and are easily recoverable by plugging it in another machine even without GlusterFS (obviously, if you pull a disk while it's being mounted some files may end up corrupt, but no worse than with single standalone disks). Of course, it also allows bundling multiple servers so that their disks appear as one to clients - that's what it's actually designed for. But offhand I can't see why it couldn't be used with a single server this way. Has anybody here used GlusterFS with Debian? Would (ab)using it that way make sense? -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: WinBlow$ Home Server equivalent
On Sat, Apr 11, 2009 at 09:11:27AM +1000, Alex Samad (a...@samad.com.au) wrote: but the marketing brief seems to suggest that if you lose a drive you only lose the files that were on there. I would take that with a big grain of salt. Marketing briefs suggesting things without explicitly saying so are generally not to be taken at face value. Unless and until someone either finds an explicit statement by M$ to that effect or testifies having actually tried that, I will remain skeptical. with lvm you would lose any lv that has blocks on that drive Not necessarily, although recovery of a partially lost lv is rather painful. And of course raid/mirroring helps - with software raid as well as with lvm mirroring you can do it even with disks of different sizes. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: boot failure diagnostics
On Sat, Apr 11, 2009 at 07:04:31AM +0200, Sven Joachim (svenj...@gmx.de) wrote: Running fsck is not necessarily a bad idea, but you should only run it on _unmounted_ filesystems. Usually fsck is also safe with read-only mounts. You can turn existing mount to read-only with mount -o remount,ro / -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: WinBlow$ Home Server equivalent
On Sat, Apr 11, 2009 at 11:28:35AM +0100, Alan Chandler (a...@chandlerfamily.org.uk) wrote: Somewhile ago after looking at all the options, I came to the conclusion that raid1 + lvm + daily cron driven backups to another machine was my best solution for keeping. That's what I do, too. Fortunately, I haven't had the need to test the full hypothosis, since todate none of my disks have failed. Lucky you. I am currently planning my next upgrades - my desktop machine has run out of disk space, my server is getting old and has also almost run out of disk space so I need to plan an upgrade path for that too. I think raid 1 will allow me to fail one disk, and then replace it with a larger one, then fail the other add replace that too. That'll work, but leaves you vulnerable to disk failure during the transition. If you can temporarily connect the new disks (or just one of them at a time) simultaneously with old ones, it'd be safer to extend the raid1 into 3-way mirror first, then reduce it back to 2-way leaving one of the old ones out, and doing that again for the second new disk. The only step I am not sure of is how to grow into the new spare space Easiest way is to partition the new disks so that first partition matches the old disks in size and create a separate raid1 pair of the second partition, and add it as new physical volume. It is also a bit safer against certain types of hardware problems. (If the new disks are big enough you might wish to partition them into even more pieces and raid them separately.) If you want to stick to one raid1 array it is of course also possible: mdadm --grow and pvresize are what you need. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Suggestions for multilevel backup of single machine?
On Sat, Apr 11, 2009 at 01:47:08PM +0100, James Youngman (j...@gnu.org) wrote: (2) It would be useful to have a historic backup capability too (e.g. the way the filesystem looked yesterday, last week, last month and a year ago), at least for filesystems like /home. What are good solutions for doing (2)? I'm using rsnapshot, and given what you told of your setup, I think you'd be happy with it, too. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: WinBlow$ Home Server equivalent
On Sun, Apr 12, 2009 at 08:15:04AM +1000, Alex Samad (a...@samad.com.au) wrote: recovery of a partially lost lv is rather painful. somebody else suggested a process on how to fix a missing pv (using /dev/zero) and then fsck'ing thus giving you a mountable partition, but how do you tell which files are valid and which are not ! Quite often you can do that by examining the files themselves, although admittedly not always. Can you outline a process that allow you to find out which pv a file exists on ? Sure: Use debugfs to determine which blocks the file uses, apply a little math to get from blocks to extents, then lvdisplay --maps to find out which pv they're on. Of course, if the lost pv contains the inode(s) of the file you want, it gets harder - even the fsck trick isn't likely to work. I have once rescued a file under such circumstances by searching for known pieces of the file content - not fun. With a binary file it would've been all but impossible. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: WinBlow$ Home Server equivalent
On Sun, Apr 12, 2009 at 03:20:38PM +1000, Alex Samad (a...@samad.com.au) wrote: i think you have just proven my point, its not a simple process Of course. As I wrote earlier: recovery of a partially lost lv is rather painful. Although if the file - pv mapping was commonly needed, it wouldn't be too hard to write a script to do it. Or the reverse, determining which files were on the lost pv. I guess there hasn't been much need since nobody's bothered to do that yet. But I have yet to see any convincing information about how much better Windows Home Server is in a similar situation. What I have seen is news headings like Windows Home Server Bug Could Lead To Data Loss and Shared Folders Problem While Removing Physically Failed Hard Drive in Windows Home Server. If someone has actual experience about dealing with lost disk in WHS, please tell us. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: WinBlow$ Home Server equivalent
On Fri, Apr 10, 2009 at 03:04:22AM -0500, Boyd Stephen Smith Jr. (b...@iguanasuicide.net) wrote: (taken from Windows Home Server Technical Brief - Drive Extender.docx) As you add more hard drives to your home server, they are treated as a single large pool of available storage space. You no longer need to deal with drive letters (such as E:, F:, and G: ) because you can add more hard drives. So far, sounds like UNIX in general or LVM in specific. Well, yes, except for the pretty GUI. All of the features you listed can be done with lvm command line tools. Which of course is not the same thing: ease of use is the whole selling point of Windows Home Server. I guess the closest Linux equivalent would be Openfiler (www.openfiler.com), but I don't know of any separate disk management GUI that could be used in Debian. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 02:17:03PM +1000, Alex Samad (a...@samad.com.au) wrote: Aiming to get a couple of 1Tb drives to migrate the 3x500Gb RAID5 array to a RAID1 and use two of the 500Gb drives for thhe new boot drive with LVM (With /boot / /home and so on on it). so you are going to have md0 = raid1 2 x 500Gb md1 = raid1 2 x 1Tb Not necessarily, it might make sense to partition the drives in smaller pieces and raid them separately. But: I would create 3 partitions on the 500GB drives 500M /boot (ext2 or ext3) 20G / (ext3) Could you explain the rationale behind this? It doesn't make any sense to me. The only (?) point in having a separate /boot is when you can't boot directly off /, like when it's in LVM or RAID5 or encrypted or something like that. If you are going to make non-LVM, non-encrypted, RAID1 ext3 /, you can boot off it directly, without separate /boot. If you are making separate /boot, you might as well put / under LVM (which is what I'd do - and indeed what I have done with just about every machine I have installed since I've forgotten when). -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Wed, Apr 08, 2009 at 06:02:26PM -0300, Henrique de Moraes Holschuh (h...@debian.org) wrote: On Wed, 08 Apr 2009, Miles Fidelman wrote: One suggestion: think very carefully about whether you really want to do this. I second that. It is really not smart to have / (or /boot) in LVM if you can help it. I suggest that a small (1GB-4GB) partition for simple md-raid1 be used for / instead. That won't give you any headaches, including on disaster recovery scenarios. I would respectfully disagree. There are significant advantages in putting / in LVM, it is a well-supported, standard configuration, and avoiding it only gives false sense of security: in a disaster situation you need to know basics of mdadm and lvm anyway, if you use them. Yes, leaving / out of LVM does give you a more complete environment to work with when system crashes in a way that LVM (the volume group containing /) is inaccessible. It doesn't help much though unless you also leave /usr out, and I've lost count on how often I've enlarged /usr and been grateful it was under LVM. All the essential tools for managing software raid and lvm are, however, available even without / - indeed they're in initrd, and if you can't use them, you're out of luck anyway. On the other hand, having / in LVM means: * you can enlarge / when necessary; * you can encrypt / if desired; * you can use other RAID configurations besides RAID1 with /; * you don't have to create separate volumes for each of /usr, /var and /tmp (although you probably should anyway); * it's the standard configuration, offered as automatic default installation option, and many people are using it so finding someone to help when needed shouldn't be hard. As for the rest of your points, well, both software raid and lvm do increase complexity and require learning some new tricks, but they're well worth the trouble if you manage any system more complex than a simple workstation, IMHO. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 06:59:48PM +1000, Alex Samad (a...@samad.com.au) wrote: I would create 3 partitions on the 500GB drives 500M /boot (ext2 or ext3) 20G / (ext3) Could you explain the rationale behind this? /boot can be loaded or and I like to have a resuce image on there just incase I stuff something up on /, for the cost of a 1 partition and a separate /boot it feel a bit better, when its part of / anything can happen to it Hmm. Are you mounting /boot read-only? That should work and it'd indeed protect it better. Otherwise the difference in safety against things like mistyped rm commands and whatnot would be marginal, I think. If you have a concrete example where separate /boot saved (or would have saved) the day, I'd be interested. As for rescue image use, maybe it would make sense in some types of use, but then I'd rather create a separate partition just for that (and leave it unmounted in normal use). But if the machine is physically easily accessible, a removable medium like USB stick would be better for that. I have had to recover to many servers and I like having root on a non lvm partition just another layer I don't need it there is a problem. That's always the tradeoff, yes. LVM gives flexibility at the cost of increased complexity, and whether it's worth the price in any given situation, well, depends. One scenario where I'd consider non-LVM root (and /usr and /var) is where the machine is located in a hard-to-get-at place, with only less experienced people present: being able to get the machine up by talking them trough the hoops up to a point where I could get in with ssh could indeed be easier then. raid1 is easy to deal with when thing go wrong Well, it has its own issues, but yes, less moving parts than in LVM (and lvm over raid1 is obviously more complex than either alone). and again the cost of it is minimal 1 partition slot and managing your data needs Cost in disk space is minimal, agreed, but there are other costs caused by the inflexibility it implies. Anyway, nowadays I usually encrypt / anyway (or rather, encrypt mdadm devices used as lvm physical volumes and put / in there), and then there's no choice. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Disk drive recovery help
On Wed, Apr 08, 2009 at 08:50:37AM -0700, tony mollica (t...@threedogs.net) wrote: fsck.ext2: No such file or directory while trying to open /dev/sdg1 What does the device file look like? Could you post 'ls -l /dev/sdg*' output? Other than a corrupt device file, it could be a hard read error in a critical place in the disk. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 09:35:57AM -0400, Douglas A. Tutty (dtu...@vianet.ca) wrote: Is there an advantage of software raid10 over multiple raid1 arrays joined with LVM? Speed. Also reduced complexity, if you can forgo LVM entirely. Disadvantages are slightly bigger danger of data loss and increased complexity, if you are using LVM anyway. Capacity can be dynamically added with pairs of disks. Yes. I'd prefer that if speed isn't critical. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 09:32:47AM -0400, Douglas A. Tutty (dtu...@vianet.ca) wrote: Yes, leaving / out of LVM does give you a more complete environment to work with when system crashes in a way that LVM (the volume group containing /) is inaccessible. It doesn't help much though unless you also leave /usr out, What is in /usr that you'd need (ok, other than man pages)? Besides man pages (which are useful if you don't have access to the Internet), there's some useful stuff in /usr/bin and /usr/sbin, like grub, mkinitramfs, ssh, find ... Nothing essential, but as I said, everything essential is on the initramdisk anyway. You don't have whatever notes you've left yourself in /root You could keep your notes in /boot. On the other hand, having / in LVM means: * you can enlarge / when necessary; You should never have to enlarge a 500 MB / Probably not, if you have separate /usr, /var, /tmp and /home, as you generally should. But it never is a long time. I've had to increase / at least twice when it was too small for OS upgrade (last in a system where it was just 50MB - which had been plenty when the box was first installed, but not anymore). * you can encrypt / if desired; Why would you need / encrypted (if swap, /tmp, /home, and parts of /var are encrypted)? To protect the notes you left in /root, of course. :-) Seriously, there is a lot of potentially sensitive information in there, like /etc/passwd, /etc/shadow, ssh keys, root's shell history, c. * you can use other RAID configurations besides RAID1 with /; True, but for 500 MB is that helpful? If you have more than 2 disks, just put a 500 MB partition on each and have more than 2 components to the raid1 array. Last I tried, booting off 3+ -way raid1 wasn't supported and didn't work (it's been a while though). So, you might want to use raid6 for reliability (or raid10 for speed, not that I can think why speed could matter in /). Of course, this point is moot if you have both /boot and / as separate, non-lvm partitions. Come to think of it, that'd allow encrypting / as well, although I can't see why that kind of non-standard setup would be better than having / in lvm. * it's the standard configuration, offered as automatic default installation option, and many people are using it so finding someone to help when needed shouldn't be hard. I've never used the automatic default; It always wastes resources on my boxes. To tell the truth, neither have I, but the point was that it is a well-known, well-supported setup. Figure out what all documentation, man pages (in text format), notes, etc that you would want Figuring that out in advance may not be that easy. and put them in /root/doc. Any scripts that you find helpful for rebuilding arrays you could put in /root/bin. You could just as well use /boot/doc and /boot/bin. But, yeah: the issue is debatable, there's no really overwhelming reason to go either way in every case. There are situations where the advantages of lvm are not important and its complexity may be a reason to avoid it. I think they're rare, but my view may be biased by the fact I've used lvm for so long that I no longer remember it ever being difficult. :-) -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 02:14:32PM -0400, Douglas A. Tutty (dtu...@vianet.ca) wrote: What load of gunk will be dumped into / to take it bigger than 500 MB? I've got a box where /lib takes 200MB now, of which /lib/modules is 140MB - and that's per kernel, during kernel updates it temporarily doubles, taking /lib to 340MB or thereabouts. I do't see it at all impossible that the 500MB I have for / there now will get too small before the machine is retired. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 11:09:15AM -0500, Boyd Stephen Smith Jr. (b...@iguanasuicide.net) wrote: Is there an advantage of software raid10 over multiple raid1 arrays joined with LVM? Speed. Not much, if any. LVM can stripe data across pvs ala RAID-0. Well, then you are doing software raid10, even though not with md. :-) But yes, the above description is ambiguous, and indeed lvm can do striping and mirroring (could before md was invented, or even before Linux ever got either). I once compared raid0 with md vs. lvm striping (being used with the latter from hp-ux), and decided to go with the former. There wasn't much difference in speed as I recall, but md allowed making bootable raid1 and it seemed better supported in Linux. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Thu, Apr 09, 2009 at 08:50:34PM -0500, Mark Allums (m...@allums.com) wrote: Douglas A. Tutty wrote: Is there an advantage of software raid10 over multiple raid1 arrays joined with LVM? Capacity can be dynamically added with pairs of disks. Only one: simplicity. It would make it easier for someone to understand, in the beginning. Curiously, I would've thought the opposite, that is, bunch of separate raid1 arrays would be easier to understand than raid10. Raid1 is conceptually simple compared to any other raid level, and if you're using lvm anyway, it doesn't make much difference whether physical volumes are disks or disk pairs. Anybody want to claim being a newbie and having an opinion here? -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Wed, Apr 08, 2009 at 03:34:57PM +0100, Kelly Harding (kelly.hard...@gmail.com) wrote: My server has got debian on it currently in the following configuration: 250Gb boot drive with partitions for / /boot /home 3x500Gb drives in RAID5 array, with XFS on top directly. Aiming to get a couple of 1Tb drives to migrate the 3x500Gb RAID5 array to a RAID1 and use two of the 500Gb drives for thhe new boot drive with LVM (With /boot / /home and so on on it). So the new configuration will be 2x1TB + 2x500GB drives in RAID1 setup and one of the 500GB drives will be removed along with the 250GB one, correct? Just wondering if theres any suggestions on how best to go about migrating like this? not really worried too much about downtime, as long as data is preserved. It would've been much easier if you'd used LVM to begin with and partitioned the drives with expansion in mind, but it shouldn't be too hard anyway. A few clarifying questions: Is there some reason against putting /boot and / on the new TB disks instead of the old ones? Can you leave all six disks connected during the transition or are you limited to four simultaneously connected disks? How full is the XFS system presently on the RAID5? -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Wed, Apr 08, 2009 at 11:57:59AM -0700, Mike Castle (dalgoda+deb...@gmail.com) wrote: On Wed, Apr 8, 2009 at 7:34 AM, Kelly Harding kelly.hard...@gmail.com wrote: Aiming to get a couple of 1Tb drives to migrate the 3x500Gb RAID5 array to a RAID1 and use two of the 500Gb drives for thhe new boot drive with LVM (With /boot / /home and so on on it). with today's larger disks, the time it takes to do a rebuild is sufficiently long enough that the risk of a second drive failure during the rebuild is high enough to be troublesome. That is true - but moving from RAID5 to RAID1 will improve the odds. If the quality of the disks is same, probability of an unrecoverable read error in one 1TB disk should be same in one of two 500GB disks, and the probability of whole disk failure is bigger with 2 disks. Moreover, RAID5 rebuild is much slower than RAID1 - I don't have hard data at hand, but I think rebuilding 3x500GB RAID5 would take longer than 2x1TB RAID1. So 2x1TB as RAID1 is safer than 3x500GB as RAID5. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Advice on raid/lvm
On Wed, Apr 08, 2009 at 06:34:09PM -0500, Mark Allums (m...@allums.com) wrote: Not really answering your question directly, but may I suggest, if cost is not *absolutely* critical, that you consider RAID 10? If it is a server, then certainly you will want to get away from a three-drive RAID 5. A RAID 10 is a good compromise between redundancy, speed, and cost. It just takes four drives instead of three (or two.) Whether it's a good compromise depends on how much speed vs. reliability matter. A 4-disk RAID10 can survive 2-disk failure with 2/3 probability (RAID 01 only with 1/3), whereas RAID6 can handle loss of any two disks but is also much slower. With RAID5 the compromise is between capacity and reliability, and as noted, the balance keeps getting worse as disks grow. It may still make sense in some scenarios though. But simply two separate RAID1 instances (which I understand OP was planning) is actually more robust than RAID10 (in that even loss of three disks won't lose _all_ data in there), and it can be done with disks of different sizes. So I'd prefer that if speed is not critical but reliability requirements don't quite warrant RAID6 either. Of course it would also be possible to divide the 1TB disks in half, use one half as RAID1 and the other as RAID10 together with the 500GB disks for speed, but if the latter are slower to begin with, the speed gain may not be all that great. So, I think OP's original plan (as I understood it) is a sound compromise between cost, reliability, speed and simplicity under a wide range of requirements. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: lvm and multiboot
On Thu, Apr 02, 2009 at 11:36:27PM -0500, zhang zhengquan (zhang.zhengq...@gmail.com) wrote: If I use lvm for the rest of the drive except /boot, can I possibly reduce the size of the lvm volume group and get some free space unformatted not controlled by lvm? You can, but that'll be a lot easier if you create multiple partitions to begin with: freeing an entire partition from lvm is easier than reducing the size of one, and it doesn't really cost anything. So, for maximum flexibility create a small (200MB or so) partition for /boot and a suitable selection of partitions with different sizes so that you can later combine them in various ways. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Debian RAM supporting.
On Wed, Apr 01, 2009 at 04:49:44PM +0500, Daniel Suleyman (danik...@gmail.com) wrote: Thank you for information i'll check it. It 1-2 years old, server is in colocation, i cant easy check it. I'm pretty sure it's some Xeon and can run 64-bit OS just fine. is there any way to check which processor installed in debian? cat /proc/cpuinfo -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Debian RAM supporting.
On Wed, Apr 01, 2009 at 05:09:30PM +0500, Daniel Suleyman (danik...@gmail.com) wrote: model name : Intel(R) Xeon(TM) CPU 3.20GHz [...] flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe lm pni monitor ds_cpl cid is it ok? for 64 bit Yes. The lm in the flags list means long mode = x86-64. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Why no Skype packages?
On Wed, Apr 01, 2009 at 02:23:21PM +0200, Dieder Vervoort (lin...@telenet.be) wrote: The problem is that a lot of windoze users are using Skype. So if you want to call them you need Skype ( if you don't have a Skype handset of course) You could use Fring (www.fring.com), it can handle Skype calls. Whether it's any better than Skype, well, depends. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: OT: eSATA card/chip opinions
On Wed, Apr 01, 2009 at 07:46:06PM -0700, Kelly Clowers (kelly.clow...@gmail.com) wrote: MASSCOOL PCI-e 4 port SATA2 (SIL3132 Chip) Model XWT-PCIE10 http://www.newegg.com/Product/Product.aspx?Item=N82E16815280008 (uses sata_sil24) I've used a couple of Sil3132-based cards and I've been happy with them. Note though that it only supports two ports - some cards (like the one you link to) have 4 ports but only two can be used simultaneously (so you can choose between external and internal ports). Also, some cheap cards use internal-style connectors externally instead of proper eSATA ones (don't know about this particular one). StarTech 2 Port PCI-Express eSATA/SATA (1 Int + 1 Ext) Card Model PEXESATA2I http://www.newegg.com/Product/Product.aspx?Item=N82E16815158095 (uses (jmicron) AHCI) No experience with this one. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Debian RAM supporting.
On Thu, Apr 02, 2009 at 11:32:26AM +0800, Bret Busby (b...@busby.net) wrote: http://www.acer.com.au/acer/product.do?link=oln23g.redirectchangedAlts=kcond5e.c2att92=853CRC=1165838950 , A 64-bit operating system is required to enjoy the ultimate performance of 4 GB memory So, does anyone who has something worth saying, know whether a 64 bit CPU and a 64 bit OS, is required to properly use = 4GB of RAM? No, for reasonable definitions of properly. But the performance of a 32-bit CPU will suffer from the extension mechanisms needed to handle memory space bigger than 2^32, so if you are buying a new machine and plan to have over 4GB of RAM, a 64-bit CPU and OS would be preferable. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Resizing LVM partition
On Sun, Mar 29, 2009 at 12:12:03PM -0300, Daniel Bareiro (daniel-lis...@gmx.net) wrote: In the end you ended up being convincing to me and I was decided to create a new physical volume that soon I added to volume group due to its rapidity and facility Good. In the procedure that you mentioned, when erasing and creating the larger partition the data that it maintainded will be lost No it won't. The contents of the partition will stay in place, fdisk only changes the partition table. When the volume group is reawakened, it finds the data where it expects, only there will be unused space at the end of the partition, and pvresize will claim it. At least that's how it works with normal disks, I don't know if your hardware RAID does something strange. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Resizing LVM partition
On Sat, Mar 28, 2009 at 12:08:43PM -0300, Daniel Bareiro (daniel-lis...@gmx.net) wrote: But I would like, to be possible, not to have to be creating partitions by each disk that could be adding. Well, it is possible, just hard compared to creating new partitions. Extra partitions don't really cost anything and make it easier to rearrange things should you wish to create multiple volume groups later. I've got a box with a total of 29 partitions on five disks, divided among four volume groups (and 2 non-lvm for soft-mirrored /boot), partly for historical reasons, partly by design, with no detectable performance impact, and I've done things like splitting a volume group in two in order to encrypt half of it, and replacing root disk without shutting the system down. I like lvm! Anyway, if you really want to enlarge the lvm partition, all you need to do is (1) use (e.g.) fdisk to remove the partition and recreate it at same starting position but new end position; (2) extend the physical volume with pvresize. You may need to boot in between, and you might wish to do (1) in single-user mode and with the volume group deactivated (vgchange -a n). Caveat emptor: any mistakes (including things I've made by forgetting something) may result in loss of data. Fresh backup before starting is recommended. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Resizing LVM partition
On Sat, Mar 28, 2009 at 12:01:18AM -0300, Daniel Bareiro (daniel-lis...@gmx.net) wrote: How I can to extend a LVM partition? Why do you want to do that? I've always thought it one of LVM's main attractions that it it makes partition resizing almost totally unnecessary. Instead, just create another LVM partition and add it to the existing volume group (pvcreate, vgextend). If there is some reason that won't do what you want, I'm curious to hear what it is. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: configure firewall in etch
On Wed, Mar 25, 2009 at 11:10:01PM -0600, Dan Schaper (dscha...@ganymeade.com)
wrote:
#!/bin/bash
ip=`ifconfig $1 | grep inet addr | awk '{print $2}' | tr -d addr:`
echo Your ip on $1 is $ip
Not that it really matters, but you don't need grep and tr:
ip=`ifconfig $1 | awk '/inet addr/{sub(/addr:/,); print $2}'`
or even shorter
ip=`ifconfig $1 | awk -F'[: ]+' '/inet addr:/{print $4}'`
and unless you need the ip in a variable for other purposes, you can
do away with the echo as well:
ifconfig $1 | awk -F'[: ]+' '/inet addr:/{print your ip on '$1' is $4}'
Also, it might be prudent to use full path for ifconfig,
i.e. /sbin/ifconfig, as /sbin may not be in $PATH.
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: [OT] Cron and day of the week
On Wed, Mar 25, 2009 at 04:25:45PM +, T o n g (mlist4sunt...@yahoo.com) wrote: I know we can put day of the week info as cron schedules, but how can I define: Cron syntax can't handle these, but with an extra test in crontab they can be done. Assuming you want to run myprog at noon: - first Monday of the month 0 12 * * 1 [ `date +%d` -le 7 ] myprog or 0 12 1-7 * * [ `date +%w` = 1 ] myprog - second Monday of the month 0 12 8-14 * * [ `date +%w` = 1 ] myprog - last Monday of the month 0 12 * * * 1 [ `date +%m` != `date --date week +%m` ] myprog The last depends on Gnu date (but that's what Debian uses, so it's appropriate here): date --date week gives the date one week from now, and if the month is different, today is the last Monday. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: about ntp
On Wed, Mar 25, 2009 at 10:08:20PM +0200, Andrei Popescu (andreimpope...@gmail.com) wrote: On Sat,25.Apr.09, 16:01:48, leo wrote: when I exec ntpdate -u pool.ntp.org for example, the out is: no server suitable for synchronization found so it's posible a misconfiguration or no access what could be the problem here? Sorry I have to ask, but is that box connected to the internet? If it weren't it wouldn't give that error but something like can't find host pool.ntp.org - obviously DNS works and it gets the IP but can't connect to the server. My bet is that either (1) port 123 is blocked on leo's end (overzealous ISP blocking all ports not explicitly requested open?) or (2) the server he gets won't talk to him (maybe because he's in Cuba and the server he happens to get is in USA and following the embargo too literally), or (3) his connection is simply so slow ntpdate gives up before getting the reply. Leo, try europe.pool.ntp.org instead, and if it fails, try again with 0.europe.pool.ntp.org, 1.europe.pool.ntp.org c up to 3. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Delete 4 million files
On Wed, Mar 25, 2009 at 03:10:42PM +, Tzafrir Cohen (tzaf...@cohens.org.il)
wrote:
$ for i in `seq 400`; do echo something quite longer; done | xargs
/bin/echo | wc
756 1200 9200
[...]
So it's indeed not 4M processes, but still quite a few.
Even 756 is much less than 4M.
But wrost:
you're traversing the directory many times. And you're telling rm in
which explicit order to remove files, rather than simply the native
order of the files in the directory (or whatever is convinient for the
implementor). Which probably requires rm a number of extra lookups in
the directory.
Interesting point; I hadn't thought of that.
How much fork() costs in comparison to reading
a directory entry, well that'd depend on things like
disk and cpu speed, available memory, filesystem type c.
To get an idea which way it falls I did a quick test with 500k files
(created by seq 50 | xargs touch) on my box.
First on an ext3 filesystem:
rm -rf testd 4m11.909s
find testd -type f |xargs rm 4m42.025s
find testd -type rm -exec rm {} \; 62m59.030s
find testd -type f -delete4m19.340s
Then on tmpfs:
rm -rf testd 0m2.507s
find testd -type f |xargs rm 0m6.318s
find testd -type rm -exec rm {} \; 58m34.645s
find testd -type f -delete 0m3.362s
So, it would seem the number of rm calls indeed dominates
the time needed, not directory traversal.
Of course here xargs was helped by the fact that filenames
were short (most 12 characters with the directory name),
but the speedup over -exec is still rather impressive.
If anyone can come up with a scenario where -exec
is significantly faster than xargs, I'd be interested.
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: Delete 4 million files
kj wrote:
Now, I've been running the usual find . -type f -exec rm {} \;
but this is going at about 700,000 per day. Would simply doing an rm
-rf on the Maildir be quicker? Or is there a better way?
While rm -rf would certainly be quicker and is obviously preferred
when you want to remove everything in the directory, the find version
could be speeded significantly by using xargs, i.e.,
find . -type f -print0 | xargs -0 rm
This is especially useful if you want to remove files selecticely
instead of everything at once.
--
Tapani Tarvainen
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: users backup (will change the root hard disk)
On Mon, Mar 23, 2009 at 04:51:19PM +0100, Pol Hallen (debenv...@fuckaround.org) wrote: Hi folks :-) I need change my root hard disk and I'd like save all system users (many because each has a mailbox). Is there a tool (or which program) can I use for do this? Are you saying you don't have a regular backup system in place? Eek. Anyway, there are lots of excellent backup programs around, but if you can connect both disks at once, you could just use the old disk as a backup and copy the stuff from there after moving system to the new disk. Better yet if you are using LVM, just use pvmove - with hotswap disks you could even replace the root disk without booting. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: which package can display chart?
On Tue, Mar 24, 2009 at 11:54:27AM +0800, Long Wind (longwind2...@gmail.com) wrote: Even after installing all security updates, can you be sure your system is safe? There may be new security problem discovered at any time True. You can never be 100% sure. But: If I do not install latest security updates and do not get attacked, am I wasting time by worrying and installing security updates? No, unless you have some magic way of ensuring you'll never get attacked. The security updates fix _known_ security problems - the bad guys will know them, too. If you don't install the patches, your system will be significantly more vulnerable than if you do. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
cryptdisks lvm2 startup sequence
I recently moved a system from Ubuntu (Hardy) to Debian (Lenny), and while most things went smoothly, external disk box with encrypted volume group did not wake up automatically. I had set things up so that keys of the encrypted volumes (7 of them) in the external box were on another (small) encrypted device, which is mounted and unmounted via /etc/default/cryptdisks during cryptdisks-early (and luksClose it in /etc/rc.local to hide the keys), in order to avoid typing seven passwords at boot. This worked fine in Hardy but failed in Lenny. Looking at the boot sequence made the reason obvious: /etc/rcS.d had S26cryptdisks-early S26lvm2 S28cryptdisks Since the volumes in question are not available until the later cryptdisks script is done, the lvm2 script doesn't see them and the volume groups in them aren't activated. I can fix this in two ways (tested, both work): (1) move the lvm2 script startup after cryptdisks, i.e., mv /etc/rcS.d/S26lvm2 /etc/rcS.d/S29lvm2 (2) redo necessary pieces of lvm2 startup after cryptdisks, e.g., add to /etc/rc.local /sbin/vgscan; /sbin/vgchange -a y; /bin/mount -a The latter has the obvious advantage that it is less likely to be broken by some upgrade later, but it is somewhat ugly. So, first question: Is there a good reason for the default order of cryptdisks-lvm2 startup? That is, is there a (real, common) situation where lvm2 has to be started up before (late) cryptdisks? And second question: is anybody else using the two-stage cryptdisk startup sequence like I do, or in some other way that opens encrypted LVM volumes only at the late cryptdisk startup script? If it's just me I guess I'm better off messing with rc.local, otherwise I'll consider making a bug report/enhancement request. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: SOLVED Re: Lenny won't install on an old Pentium that used to run Etch. Try 2
On Sat, 21 Mar 2009 16:50:32 -0400, Stefan Monnier posted: FWIW, I find that running Lenny on a 64MB machine is bearable but slowish, and Etch was already too slow on a 32MB machine. So make sure you have enough RAM. For very simple uses, it works with less RAM, but as soon as you need to apt-get install or apt-get upgrade the memory limit becomes severe. Yeah. You may have to shut every unnecessary process down just to run aptitude update. But Lenny isn't too bad. I installed Lenny on a 64MB box a while back, and first it appeared almost unusable, but after a bit of tweaking it's now running happily as a DNS server and mail relay. The biggest memory hog was rsyslogd, but it's memory footprint shrunk from about 28MB down to 4MB simply by adding 'ulimit -s 128' to /etc/default/rsyslog. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
BUG: ... kernel NULL pointer reference ... udevadm timeout
00 00 00 89 c3 75 25 e8 73 ff ff ff ba f0 ff ff ff [ 31.144520] EIP: [c021bca5] pnp_activate_dev+0x3/0x3a SS:ESP 0068:c3edfdd8 [ 31.155461] ---[ end trace 029a775029b931f2 ]--- [ 32.877340] 3c59x :01:00.0: enabling device ( - 0003) [ 32.877510] 3c59x: Donald Becker and others. [ 32.877635] :01:00.0: 3Com PCI 3CCFE575CT Tornado CardBus at ca9c8000. [ 32.877759] PCI: Setting latency timer of device :01:00.0 to 64 Any suggestions would be appreciated. -- Tapani Tarvainen -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
