re: Fixing vinum after removing a HDD
Date: Thu, 16 Dec 2004 21:56:29 -0800 (PST) From: orville weyrich [EMAIL PROTECTED] Subject: re: Fixing vinum after removing a HDD To: [EMAIL PROTECTED] On Wed, 2004-02-11 at 17:02, Greg 'groggy' Lehey wrote: resetconfig The resetconfig command completely obliterates the vinum configu- ration on a system. Use this command only when you want to com- pletely delete the configuration. I'm completely baffled that people want to use this command so much. The correct sequence is to remove the drive, replace it with something else with a Vinum partition, and then start the defective objects. It depends on the volume structures as to whether the data can still be recovered. PLEASE PLEASE PLEASE elaborate on what volume structuers CAN be recovered after resetconfig and what strucuters cannot. I am trying to deal with *invalid* drives in dumpconfig and not getting any answers. Thanks __ Do you Yahoo!? Yahoo! Mail - You care about security. So do we. http://promotions.yahoo.com/new_mail __ Do you Yahoo!? Dress up your holiday email, Hollywood style. Learn more. http://celebrity.mail.yahoo.com ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Has anybody EVER successfully recovered VINUM?
See below --- Greg 'groggy' Lehey [EMAIL PROTECTED] wrote: create original.config.file Yes, that's basically it. I tried that already. It did not have the desired effect. It created TWO MORE plexes, that did not have disk space to back them up (because the drives were fully allocated by the original configuration. Correct. This isn't the way to do it. I've just posted the correct URL: http://www.vinumvm.org/vinum/replacing-drive.html OK -- I have gotten to the point where I have a valid plex (I did it by installing a large, cheap IDE drive in the system and putting an entirely new plex on it). But when I try to mount the drive, mount complains that the volume needs to be fscked. Fsck refuses. I CAN however mount the file system and see the expected top level directories (liberty mysql orville reform tam). But I cannot view contents inside the direcotries. (see selected transcript below). So is my data toast, or is there a way to fsck /dev/vinum/raid? SELECTED TRANSCRIPT: S raid.p3.s0State: up PO:0 B Size: 2151 MB S raid.p3.s1State: up PO: 512 kB Size: 2151 MB S raid.p3.s2State: up PO: 1024 kB Size: 2151 MB S raid.p3.s3State: up PO: 1536 kB Size: 2151 MB S raid.p3.s4State: up PO: 2048 kB Size: 2151 MB S raid.p3.s5State: up PO: 2560 kB Size: 2151 MB S raid.p3.s6State: up PO: 3072 kB Size: 2151 MB S raid.p3.s7State: up PO: 3584 kB Size: 2151 MB S raid.p3.s8State: up PO: 4096 kB Size: 2151 MB S raid.p3.s9State: up PO: 4608 kB Size: 2151 MB vinum - quit bashful# mount /dev/vinum/raid /raid WARNING: R/W mount of /raid denied. Filesystem is not clean - run fsck mount: /dev/vinum/raid: Operation not permitted bashful# fsck /dev/vinum/raid ** /dev/vinum/raid BAD SUPER BLOCK: VALUES IN SUPER BLOCK DISAGREE WITH THOSE IN FIRST ALTERNATE /dev/vinum/raid: NOT LABELED AS A BSD FILE SYSTEM (unused) bashful# mount -r /dev/vinum/raid /raid bashful# ls /raid liberty mysql orville reform tam bashful# ls /raid/liberty ls: /raid/liberty: Bad file descriptor bashful# ls /raid/reform /raid/reform bashful# ls /raid/reform/* ls: No match. bashful# ls /raid/orville ls: /raid/orville: Bad file descriptor __ Do you Yahoo!? Send holiday email and support a worthy cause. Do good. http://celebrity.mail.yahoo.com ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Has anybody EVER successfully recovered VINUM?
See below --- Toomas Aas [EMAIL PROTECTED] wrote: I once had a problem with *invalid* vinum drive which was solved following this advice from Greg Lehey: http://makeashorterlink.com/?V29425BF9 So the procedure that worked for you was??? resetconfig create original.config.file I have seen alusions to this procedure, and was planning to try it as a last resort (last because of the dire warnings that are associated with resetconfig). Nowhere however have I seen the above two lines explicitly juxtaposed, so I have been reluctant to try it. Before I try the above, I had an idea last night that I want to try: I will buy a large, cheap IDE drive (big enough to hold an entire plex) and ADD it to my configuration (with all subdisks on the IDE drive). At this point, I expect to be able to sync all subdisks of the IDE plex against one or the other of the existing plexi to obtain a completely functional plex, at which point my RAID should come alive for me to back it up. Does this plan seem workable? __ Do you Yahoo!? Yahoo! Mail - Find what you need with new enhanced search. http://info.mail.yahoo.com/mail_250 ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Has anybody EVER successfully recovered VINUM?
See below --- Greg 'groggy' Lehey [EMAIL PROTECTED] wrote: On Thursday, 9 December 2004 at 7:06:35 -0800, --- Toomas Aas [EMAIL PROTECTED] wrote: I once had a problem with *invalid* vinum drive So the procedure that worked for you was??? resetconfig This doesn't sound like what you want to do. Which part of http://www.vinumvm.org/vinum/replacing-drive.html is a problem for you? The problem is that I have two subdisks, raid.p0.s9 and raid.p1.s4 that originally were associated with the drive ahc0t15. But when I try to add a new drive with the name ahc0t15, it does not associate itself with the two subdisks. Trying to start the subdisks gives the error: vinum - start raid.p0.s9 Can't start raid.p0.s9: Drive is down (5) Dumpconfig shows that the drive associated with the two subdisks is *invalid*: sd name raid.p0.s9 drive *invalid* plex raid.p0 len 4402583s driveoffset 265s state crashed plexoffset 0s but also shows the unused drive: Drive ahc0t15: Device /dev/da9s1e Created on bashful.weyrich.com at Wed Dec 8 17:24:07 2004 Config last updated Wed Dec 8 17:37:27 2004 Size: 4512230400 bytes (4303 MB) The procedure replacing-drive.html did work in another case. It doesn't seem to work here. Is there some other way to tell vinum that drive ahc0t15 should be associated with the subdisks raid.p0.s9 and raid.p1.s4? __ Do you Yahoo!? Take Yahoo! Mail with you! Get it on your mobile phone. http://mobile.yahoo.com/maildemo ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Has anybody EVER successfully recovered VINUM?
see below --- Toomas Aas [EMAIL PROTECTED] wrote: orville weyrich wrote: So the procedure that worked for you was??? resetconfig NO!!! At no point in the procedure did I do resetconfig. create original.config.file Yes, that's basically it. I tried that already. It did not have the desired effect. It created TWO MORE plexes, that did not have disk space to back them up (because the drives were fully allocated by the original configuration. It took me a while to figure out how to undo that (I had to stop plexi p0 and p1 before trying to stop and remove the new plexi p2 and p3). __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Has anybody EVER successfully recovered VINUM?
See below --- Toomas Aas [EMAIL PROTECTED] wrote: orville weyrich wrote: In any event, the example shown in the link lacks context. Is the failed drive named data1? It is automagically replaced by a new drive named data3? In all cases I've replaced a failed drive, I have named the new drive the same as old drive. THANK YOU THANK YOU THANK YOU That solved HALF my problem (i.e, it allowed me to revive one of my two down disks. This also shows that vinum CAN handle a fault in both plexes at the same time. BUT BUT BUT I still have another problem -- the same trick does not work for the OTHER down disk. vinum-dumpconfig shows: sd name raid.p0.s9 drive *invalid* plex raid.p0 len 4402583s driveoffset 265s state crashed plexoffset 0s Note the invalid where the name of my disk should be. Apparently VINUM has forgotten that raid.p0.s9 should be associated with the disk named ahc0t15, so creating a new disk named ahc0t15 does not associate with the necessary subdisk. (*nvalid* does not work as a disk name). How do I deal with THIS issue? __ Do you Yahoo!? Yahoo! Mail - 250MB free storage. Do more. Manage less. http://info.mail.yahoo.com/mail_250 ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Has anybody EVER successfully recovered VINUM?
I have been trying to figure out how to get VINUM to recognize a new disk after a disk failure, and no luck at all. I cannot find instructions in the official documentation, nor in the FreeBSD Dairy. Lots of places tell how to build a VINUM system. Nobody ever takls about how to recover from a disk failure. Can someone PLEASE help me recover? I have already posted complete information to this list, with no answer. I will give a short version now and provide more info if requested. I am running FreeBSD 4.10. My current vinum list includes the following: S raid.p0.s2State: up PO: 1024 kB Size: 2149 MB S raid.p0.s3State: up PO: 1536 kB Size: 2149 MB S raid.p0.s4State: up PO: 2048 kB Size: 2149 MB S raid.p0.s5State: up PO: 2560 kB Size: 2149 MB S raid.p0.s6State: up PO: 3072 kB Size: 2149 MB S raid.p0.s7State: up PO: 3584 kB Size: 2149 MB S raid.p0.s8State: up PO: 4096 kB Size: 2149 MB S raid.p0.s9State: crashed PO:0 B Size: 2150 MB S raid.p1.s0State: up PO:0 B Size: 2151 MB S raid.p1.s1State: up PO: 512 kB Size: 2151 MB S raid.p1.s2State: up PO: 1024 kB Size: 2151 MB S raid.p1.s3State: up PO: 1536 kB Size: 2151 MB S raid.p1.s4State: obsolete (detached) Size: 2150 MB S raid.p1.s5State: reborn PO: 2560 kB Size: 2151 MB S raid.p1.s6State: up PO: 3072 kB Size: 2151 MB S raid.p1.s7State: up PO: 3584 kB Size: 2151 MB S raid.p1.s8State: up PO: 4096 kB Size: 2151 MB S raid.p1.s9State: up PO: 4608 kB Size: 2151 MB S raid.p2.s0State: stalePO: 2304 MB Size: 2150 MB The above represents a total of 10 drives, in a striped raid configuration (half of each disk in each plex). Subdisks p0.s0 and p1.s5 are on one failed disk, Subdisks p0.s9 and p1.s4 are on a second failed disk Subdisks raid.p2.s0 and raid.p2.s1 are on the replacement disk that I was trying to install. I tried detaching the replacement subdisk and failed subdisk, and then reattaching the replacement in the failed position, and made things worse. Can somebody PLEASE tell me two things: (1) What sequence of steps SHOULD I have taken to replace the disks (I promise I will test it and document it for all to see). (2) How can I recover NOW? (I seem to recall reading somewhere that it is actually possible to reset the config and recreate it with the same config file as used originally without destroying any data, and then judicially use setstate to mark the valid subdisks as up and the invalid ones as obsolete. But this is a drastic step that I don't want to take without some guidance. Please (grovel grovel) help me! Thanks orville. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
VINUM Disaster Recovery
I have a 10 disk VINUM configuration and two of the disks are trashed. In theory there is still enough redundant information to get things working again without data loss. Vinum has detected a configuration error (duh -- two disks are toast, plus in recovery I accidently created two more plexes) and taken upon itself to stop configuration updates to prevent any further corruption (thanks! :-). At this point I have looked at http://www.vinumvm.org/vinum/how-to-debug.html and have run a command like the following: ( dd if=/dev/da9s1e skip=8 count=50 | tr -d '\000-\011\200-\377' ; echo ) da9s1e.log on all 10 disks to obtain a file containing each disk's on-disk configuration. As hoped, eight of the disks show an output similar to the attached file da1s1e.log (differing only as expected in the first line). See attached flog file for a sample output. PLEASE HELP CONFIRM MY PLAN (FOLLOWING) FOR PROCEEDING -- I DO NOT WANT TO DO ANYTHING DISASTEROUS. My thought is that I need to turn on updates, then delete the two unwanted plexes raid.p2 and raid.p3(which were accidentally created), detach the corrupt sdisks, and then hopefully VINUM will forget about the two disks that are toast (or do I somehow have to tell VINUM to forget the disks?). My plan is as follows: First, selectively start vinum: vinum - read /dev/da1s1e /dev/da2s1e /dev/da3s1e /dev/da4s1e /dev/da5s1e /dev/da6s1e /dev/da7s1e /dev/da8s1e Second, enable configuration updates: vinum-setdaemon 0 Third, save the configuration: vinum-saveconfig Fourth, stop and remove the two unwanted plexes and all attached subdisks: vinum-stop -f raid.p3 vinum-stop -f raid.p2 vinum-rm -r raid.p3 vinum-rm -r raid.p2 Fifth, stop and detach the corruped subdisks: vinum-stop -f raid.p0.s0 vinum-stop -f raid.p0.s9 vinum-stop -f raid.p1.s4 vinum-stop -f raid.p1.s5 vinum-detach raid.p0.s0 vinum-detach raid.p0.s9 vinum-detach raid.p1.s4 vinum-detach raid.p1.s5 At this point I expect to have a functional volume that can be mounted and backed up, prior to the next step of reinstalling the crashed disks, creating new subdisks, attaching them to the plexes, and resynching. PLEASE CONFIRM MY APPROACH OR TELL ME WHERE I AM WRONG! Thanks orville __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com da1s1e.log Description: da1s1e.log ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Sendmail mungs sender address
I was running a FreeBSD 4.3 system. It crashed and I built up a new FreeBSD 4.10 system. But I cannot get Sendmail to work the way it did before. The relevant part of my network includes two hosts, call them FIRE and BASHFUL. FIRE is my firewall machine, It runs FreeBSD 4.3 and has not changed. It relays mail from my internal network to the world and back. BASHFUL is behind the firewall, is my new 4.10 machine, and is configured to use FIRE as its smart host. FIRE is known to the universe as FIRE.MYDOMAIN.COM. I want mail sent from BASHFUL to carry a sender address of [EMAIL PROTECTED] Instead it carries a sender address of [EMAIL PROTECTED] (if sent from pine on BASHFUL) or [EMAIL PROTECTED] (if sent from a command-line invocation of sendmail itself). All changes to sendmail.cf and pine configuration seem to be ignored). What has changed that my outgoing mail is broken? My incoming mail works fine. Thanks Orville Weyrich __ Do you Yahoo!? Yahoo! Mail - now with 250MB free storage. Learn more. http://info.mail.yahoo.com/mail_250 ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Problems with VINUM recovery
I was trying to test my ability to recover from a disk crash and managed to toast my system disk. Therefore, I do not have any relevant vinum history file or messages file. What happened is that I purposely blew away the drive named ahc0t15, and was trying to recover it. In error, I ran the create command A SECOND TIME with the exact same configuration file as the first time, and it created two aditional plexes, which it of course couldn't fit onto the physical disks, except for raid.p2.s0, raid.p2.s9, raid.p3.s5 and raid.p3.s5. I decided that it wasn't working properly, and was in the process of zapping my entire vinum volume to recreate it and try again (I had decided that my plex pattern was not the best), and had zapped the drive ahc0t02, when I accidently blew away the system disk, crashing my system. The good news is that most of the important data I care about on the system disk was copied to the vinum volume just to put some data on the volume -- but now I really WANT that data on the vinum volume, because it is the most recent backup. As it stands now, I want to: (1) delete raid.p2 and raid.p3 (2) rebuild drive ahc0t02 to recieve revived raid.p0.s0 and raid.p1.s5 (3) rebuild drive ahc0t15 to receive revived raid.p0.s9 and raid.p1.s4 (4) revive raid.p0.s0 from the valid raid.p1.s0 (5) revive raid.p0.9 from the valid raid.p1.s9 (6) revive raid.p1.s4 from the valid raid.p0.s4 (7) revive raid.p1.s5 from the valid raid.p0.s5 I think (hope!) all of this is possible. My vinum volume was created on a vanilla FreeBSD 4.3 system. The system has now been reloaded with a FreeBSD 4.10 system in order to produce the vinum list output attached (let me know if you have trouble reading the file as attached). Trying to rm raid.p2.s8 gives the message Cant remove raid.p2.s8: Device busy (16) *** Warning configuration updates are disabled. *** I am afraid to reenable configuration updates until I am sure I know what I am doing. Since I messed up in my previous attempts, and I am now between a rock and a hard place, I need some guidance regarding how to recover -- the failed drill has suddenly become real :-( I have looked at the document http://www.vinumvm.org/vinum/replacing-drive.html and am not sure how to apply it to the above scenario. Please please help ... please orville weyrich VINUM LISTING === 8 drives: D ahc0t03 State: up Device /dev/da1s1e Avail: 2152/4303 MB (50%) D ahc0t04 State: up Device /dev/da2s1e Avail: 2152/4303 MB (50%) D ahc0t09 State: up Device /dev/da3s1e Avail: 2152/4303 MB (50%) D ahc0t10 State: up Device /dev/da4s1e Avail: 2152/4303 MB (50%) D ahc0t11 State: up Device /dev/da5s1e Avail: 2152/4303 MB (50%) D ahc0t12 State: up Device /dev/da6s1e Avail: 2152/4303 MB (50%) D ahc0t13 State: up Device /dev/da7s1e Avail: 2152/4303 MB (50%) D ahc0t14 State: up Device /dev/da8s1e Avail: 2152/4303 MB (50%) D ahc0t02 State: referenced Device Avail: 0/0 MB D *invalid* State: referenced Device Avail: 0/0 MB D ahc0t15 State: referenced Device Avail: 0/0 MB 1 volumes: V raid State: up Plexes: 4 Size: 21 GB 4 plexes: P raid.p0 S State: corrupt Subdisks:10 Size: 21 GB P raid.p1 S State: corrupt Subdisks:10 Size: 21 GB P raid.p2 S State: faulty Subdisks:10 Size: 21 GB P raid.p3 S State: corrupt Subdisks:10 Size: 21 GB 40 subdisks: S raid.p0.s0State: crashed PO:0 B Size: 2151 MB S raid.p0.s1State: up PO: 512 kB Size: 2151 MB S raid.p0.s2State: up PO: 1024 kB Size: 2151 MB S raid.p0.s3State: up PO: 1536 kB Size: 2151 MB S raid.p0.s4State: up PO: 2048 kB Size: 2151 MB S raid.p0.s5State: up PO: 2560 kB Size: 2151 MB S raid.p0.s6State: up PO: 3072 kB Size: 2151 MB S raid.p0.s7State: up PO: 3584 kB Size: 2151 MB S raid.p0.s8State: up PO: 4096 kB Size: 2151 MB S raid.p0.s9State: crashed PO: 4608 kB Size: 2151 MB S raid.p1.s0State: up PO:0 B Size: 2151 MB S raid.p1.s1State: up PO: 512 kB Size: 2151 MB S raid.p1.s2State: up PO: 1024 kB Size: 2151 MB S raid.p1.s3State: up PO: 1536 kB Size: 2151 MB S raid.p1.s4State: obsolete PO: 2048 kB Size: 2151 MB S raid.p1.s5State: crashed PO: 2560 kB Size: 2151 MB S raid.p1.s6State: up
FDISK Woes
I did a stupid thing and zapped my system FreeBSD 4.3 system disk (/dev/ad0) with FDISK and Label (in /stand/sysinstall)instead of zapping /dev/da0 that I intended to do. Needless to say, my system got sick. I have built a new FreeBSD 4.10 system on another disk and have the old (zapped) system disk installed in the system (but not mounted). Is there a way to reconstruct the disk label? (I need to find the locations of the / (ad0s1a), /usr (ad0s1f), and /var (ad0s1e), and then write a suitable disk label without destroying the data on the disk. I feel so dumb doing this, but it seems strange that FDISK would allow me to zap the label of a mounted disk. Anyway, please help me save my system! TIA orvile. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Sendmail mungs sender address SOLVED
Thanks to those who replied. I solved the problem by building a new sendmail.cf file with the following added features not found in the stock freebsd.cf: define(`SMART_HOST',`FIRE') FEATURE(accept_unresolvable_domains) MASQUERADE_AS(`MYDOMAIN.COM') and tweeking my /etc/hosts file on BASHFUL --- orville weyrich [EMAIL PROTECTED] wrote: I was running a FreeBSD 4.3 system. It crashed and I built up a new FreeBSD 4.10 system. But I cannot get Sendmail to work the way it did before. The relevant part of my network includes two hosts, call them FIRE and BASHFUL. FIRE is my firewall machine, It runs FreeBSD 4.3 and has not changed. It relays mail from my internal network to the world and back. BASHFUL is behind the firewall, is my new 4.10 machine, and is configured to use FIRE as its smart host. FIRE is known to the universe as FIRE.MYDOMAIN.COM. I want mail sent from BASHFUL to carry a sender address of [EMAIL PROTECTED] Instead it carries a sender address of [EMAIL PROTECTED] (if sent from pine on BASHFUL) or [EMAIL PROTECTED] (if sent from a command-line invocation of sendmail itself). All changes to sendmail.cf and pine configuration seem to be ignored). What has changed that my outgoing mail is broken? My incoming mail works fine. Thanks Orville Weyrich __ Do you Yahoo!? Yahoo! Mail - now with 250MB free storage. Learn more. http://info.mail.yahoo.com/mail_250 ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED] __ Do you Yahoo!? Read only the mail you want - Yahoo! Mail SpamGuard. http://promotions.yahoo.com/new_mail ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]