GEOM issue -- making changes to slices/partitions on used disk
This happened on a FreeBSD 6.0 machine. Installed a SCSI drive on the machine -- drive had been used for testing previously and had an ext2 FS on it. I was able to use the sysinstall fdisk utility to remove the ext2fs partition and create a FreeBSD slice that utilized the entire disk. Then I loaded up the sysinstall disklabel utility, setup my partitions and when I went to write my changes, recieved an error about being unable to mount these new partitions -- said mount point did not exist. I decided I'd just try it again, so I exited sysinstall and restarted it. In the disklabel tool, I added only a swap partition. Swap partition was created without a problem. Exited sysinstall, restarted it to add the other new partitions. I could not write anything, however, as whenever I attempted to do so, I got the error message ERROR: Unable to write data to disk da1. I decided I'd remove the slice and start fresh, however when I attempted to write this change, I again got an error message stating that data could not be written. At this point I did some research, and it seems that the issue is related to GEOM not allowing alteration of MBR/slices/partitions on used devices, and that the solution is to set sysctl knob kern.geom.debugflags=16 (Found it on this thread: http://groups.google.com/group/mailing.freebsd.current/browse_thread/thread/e7db21985d1b1cc3/3113b6502c5450ef?lnk=stq=freebsd+sysinstall+fdisk+ERROR%3A+Unable+to+write+data+to+diskrnum=4hl=en#3113b6502c5450ef) Also on this thread is a word of caution, stating that use of this sysctl knob has been known to result in dev entries getting confused and blowing away the wrong partition -- which would obviously be quite undesirable. This seems like a rather clunky solution, and the posts in the cited thread are two years old, so I am wondering if there's another way to go now, or if the bugs are now worked out and it's safe to use this sysctl knob. One more thing to note: the posts in the thread mention 5.x specifically, but the behavior certainly seems to match that encountered on my 6.0 box. Any thoughts or input are quite appreciated! Thank you, John -- [EMAIL PROTECTED] John Fox, Systems Administrator InfoStructure - http://www.mind.net Vox: (541)773-5000 / Fax: (541)488-7599 ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Status of 6.0 for production systems
I remember back a while when 5.x had been recently released as STABLE and the conventional wisdom said not to use it in production until the 5.3 release. Is there any such conventional wisdom as regards 6.x? Thanks, John -- [EMAIL PROTECTED] John Fox, Senior Systems Administrator InfoStructure - http://www.mind.net Vox: (541)773-5000 / Fax: (541)488-7599 ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
DNSBL - including descriptive text in sendmail error message
I am configuring a mail server to make use of the DNS-based black lists supplied by spamhaus.org, and have run into a bit of a snag. I'm running sendmail 8.12.11, and have successfully configured the 'dnsbl' feature to reject mail from hosts in the black lists, but spamhaus.org has requested that MTA error messages contain the descriptive text corresponding to the blacklisted host (e.g. http://www.spamhaus.org/SBL/sbl.lasso?query=SBL14030;). I've checked sendmail docs for information on how to do this, and come up with nothing. I've also looked for the information on the spamhaus site itself, but to no avail. Likewise for Google searches. The black lists themselves are quite helpful, and so I would like to be in compliance with spamhaus' request. Any clues would be much appreciated. BTW, I hope it's not inappropriate to have posted this to both -isp and -questions. Please let me know if I should not have done so. Thank you much, -John -- +---+ | John Fox jjf @ mind.net| System Administrator | InfoStructure | +---+ | I used to trust the media to tell me the truth, tell us the truth | | But now I've seen the payoffs everywhere I look | | Who can you trust when everyone's a crook?| | -- Queensryche, Revolution Calling | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
dumps freeze when invoked by amanda's 'sendbackup'
Hello, I've found some strange behavior on one of our FreeBSD 2.2.2 machines, and am hoping that someone here may be able to shed some light on it for us. As mentioned, the machine is running FreeBSD 2.2.2 and AMANDA 2.4.0. A while back, we noticed that dumps of the machines '/usr' partition seemed to freeze as soon as they were started; the 'dump' processes appeared in 'ps' output, but they stayed there all day long (our backups typically finish by around 10:00 AM at the latest). At the same time, dmesg showed us hard errors reading from /dev/sd0s1g (the partion holding /usr). We eventually killed the dumps. As a test, we then tried a manual dump. I was able to successfully dump that same file system over the network to a drive on another machine. It seemed wierd that a manual dump went fine, but that the dumps spawned by 'sendbackup' did not go at all. However, as I mentioned, we'd seen that drive error, so we removed all of that machine's sd0 partitions from amanda's disklist (system has two other drives, and backups went fine for them) until we could get the drive replaced, which we did last night. Replacement of drive went just fine, and once we'd verified everything was up and running properly, I edited the disklist file and re-enabled backups of the machine's '/usr'. I got in this morning and found that again, the dumps are frozen. 'dmesg' shows nothing except the boot-up messages. I'm rather frustrated at this point, trying to understand how this could be happening. I have checked permissions on the device files, and 'bin', the user amanda runs as on that machine, is a member of the operator group, which has read access to the device files, so it doesn't seem a permissions problem. I doubt it's a drive problem, because the same behavior on two different drives by two different manufacturers? I suppose it's possible, but it seems unlikely. I should mention as well that all drives on system are SCSI 2 50 pin format, under the control of an Adaptec 2940 controller card. So if it's not permissions, and not the drive (although I realize that I haven't really ruled either of these entirely out) then might it be the controller? But if it's the controller, why no problems with other drives in system? Any thoughts would be most welcome. -John -- +---+ | John Fox jjf @ mind.net |System Administrator | InfoStructure | +---+ | I used to trust the media to tell me the truth, tell us the truth | | But now I've seen the payoffs everywhere I look | | Who can you trust when everyone's a crook?| | -- Queensryche, Revolution Calling | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Viability of 5.X line for production use
Hello, We're planning some new mail servers, and until now I've been assuming that they would run under FreeBSD 4.X. But it occurred to me this morning that the 4.X line is going to go away relatively soon, and that perhaps I'd be better off going with 5.X for these new boxen, as it would probably simplify the upgrade (keeping up with releases and bug fixes) process. So I'm wondering -- do the experts here judge 5.X as ready for use in a production environment, or would that be asking for trouble? Any thoughts appreciated, John -- +---+ | John Fox [EMAIL PROTECTED] | System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
kernel option to increase max open files
Hi, folks. I've got a 2.2.2 machine which I can't currently upgrade, due to funding issues. This machine handles a lot of mail, and I am lately finding dmesg output to be full of file: table is full messages, and entries like these in the sendmail log: SYSERR(UID0): readqf: cannot open dfAAA29006: Too many open files in \ system SYSERR(UID0): Cannot open hash database /etc/aliases: Too many open \ files in system So I figure I need to increase the max number of open files available. I've looked around, and it appears I have two different options: 1) Kernel option options OPEN_MAX=### which sets the max number of open files per user (which would be 'root' in this case, as that's who sendmail runs as). 2) a login.conf class setting assigned to root. The 'openfiles' setting seems somewhat equivalent to the OPEN_MAX option. However, I am not certain that limits set in login.conf apply to a process being run by a user that didn't log in (sendmail being started by init at boot, rather than from root's login session). I welcome any thoughts/advice on this matter, as I don't want to try either of these changes without having done due diligence. Thank you, -John -- +---+ | John Fox [EMAIL PROTECTED] | System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
max IDE drive capacity supported in 2.2.2 release
Hello to all, I am trying to determine the maximum (IDE) drive capacity supported by FreeBSD 2.2.2. I have carefully read the release notes for 2.0 and 2.2.2 and seen no mention of this datum. At this point, I'm theorizing that there the only such limit has to do with the limitations of what the BIOS will support. Any thoughts appreciated, -John -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
difficulties replacing bad libssl.so.3 with newly build libssl.so.3
Hello, all. I've got a machine running FreeBSD 4.8-RELEASE, and I need to upgrade it to fix the SSL, procfs issues that have come up lately. The machine's root partition lacks the space to accomodate world source and object files, so I have two symlinks for /usr/src and /usr/obj: /usr/src -- /usr/local/world_src /usr/obj -- /usr/local/world_obj I cvsupped RELENG_4_8 and it was all placed under /usr/local/world_src as I'd expected, and cd'ing into '/usr/src' I can see the same stuff as if I'd cd'd into /usr/local/world_src. No surprises there. (In the supfile, I have *default prefix=/usr. Dunno if that applies to this discussion.) I cd'd to /usr/src and issued the buildworld command. The build ran for a while and then finished. I took a look at the contents of '/usr/obj' (aka /usr/local/world_obj) and saw a directory hierarchy that matched my '/usr/local/world_src' path: [vmail:/usr/local/etc]$ ls -lR /usr/obj/|head -n 14 total 4 drwxr-xr-x 3 root wheel 512 Oct 8 13:38 usr/ /usr/obj/usr: total 4 drwxr-xr-x 3 root wheel 512 Oct 8 13:38 local/ /usr/obj/usr/local: total 4 drwxr-xr-x 16 root wheel 512 Oct 8 13:42 world_src/ /usr/obj/usr/local/world_src: total 64 drwxr-xr-x 32 root wheel 512 Oct 8 13:41 bin/ This struck me as wierd, but I figure it's the way it ought to be, so I moved on. One of my aims was to replace libssl.so.3 with a fixed version, so (after making a backup copy of the current /usr/lib/libssl.so.3) I placed /usr/obj/usr/local/world_src/secure/lib/libssl/libssl.so.3 into /usr/lib and then attempted an https connection to the server. (Apache's mod_ssl was dynamically linked against libssl.so.3). I found that my connection did not really work properly, creating errors such as these in the httpd error log: [Wed Oct 8 16:01:04 2003] [error] [client W.X.Y.Z] Invalid method in request \x80C\x01\x03 [Wed Oct 8 16:02:48 2003] [error] [client W.X.Y.Z] Invalid method in request \x16\x03 [Wed Oct 8 16:02:48 2003] [error] [client W.X.Y.Z] Invalid method in request \x16\x03 Clearly, I did something wrong, for when I replaced the prior libssl.so.3, my errors went away. I'm trying to understand now if the root problem is: 1) My '/usr/src' and '/usr/obj' symlinks. I can't see them being the problem, but I want leave no stone unturned. 2) Incorrect technique used to replace the libssl.so.3 library. (I am leery of running 'installworld' on this machine, as it is a production system, and I've never built the world on a production machine.) 3) Something else entirely. Any clues or pointers as to the cause of the problem would be greatly appreciated. With thanks, John -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
vinum crashed disk error
Hello, Three IDE drives multiplexed together to make one large partition (for mounting as /usr/local). We were messing with hardware in the box and when we rebooted vinum spat out errors about defective objects and the boot came to a halt. We figured we had left something loose or unplugged on the motherboard, so we shut it down and took a look. Sure enough, the plug in the mobo's secondary IDE channel was loose, so we reseated it and powered the machine up again. We saw that the kernel found all the IDE drives and figured the problem was over. But vinum had the same problem. It said (loose quotation): /dev/ mounted read-only. vinum config not being rebuilt. And then spit out the same errors it had the first time. Unfortunately, the vinum.org domain is having problems of some sort, and the vinum help pages on lemis.com are redirected to the vinum.org site, so I am deprived of a great trouble-shooting resource. System is 4.8-STABLE. Below you will see the output of 'vinum start'. Any suggestions as to fixing this problem would be greatly appreciated, as we do make production use of this box. Thank you in advance for any words you may be able to offer. -John output of 'vinum start': Warning: defective objects V bigdisk State: down Plexes: 1 Size: 23 GB P big_plexC State: faulty Subdisks: 3 Size: 23 GB S drive0State: down PO:0 B Size: 8063 MB S drive2State: crashed PO: 8063 MB Size: 8063 MB S drive3State: crashed PO: 15 GB Size: 8063 MB -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
vinum crashed disk error -- addendum
Maybe it'll help if I give some more comprehensive information: vinum - l 3 drives: D ide0e State: up Device /dev/ad0s1e Avail: 0/8063 MB (0%) D ide2e State: up Device /dev/ad2s1e Avail: 0/4031 MB (0%) D ide3e State: up Device /dev/ad3s1e Avail: 0/8063 MB (0%) 1 volumes: V bigdisk State: down Plexes: 1 Size: 23 GB 1 plexes: P big_plexC State: faulty Subdisks: 3 Size: 23 GB 3 subdisks: S drive0State: down PO:0 B Size: 8063 MB S drive2State: crashed PO: 8063 MB Size: 8063 MB S drive3State: crashed PO: 15 GB Size: 8063 MB Thanks, John -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ |Gideon: I thought you said don't hold a grudge.| | Galen: I don't. I have no surviving enemies...at all. | | -- Crusdade, _Racing the Night_ | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Huge load average on mail server
I recently became responsible for our mail server, and lately, the load average on the thing has been running consistently high, in mid-20's to mid-30's. And at the same time, sendmail has become extremely sluggish in terms of responding to connections on port 25. There's got to be a bottleneck somewhere, but my experience leaves me ill-equipped to seek it out. What procedures would folks recommend for me to follow in order to try and determine the problem? Thank you, john -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ | The people and friends that we have lost, the dreams that have faded... | | never forget them. -- Yuna, Final Fantasy X| +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Free Block Count Not Updating Until After Reboot
Hello, all. I am running FreeBSD 4.8-RELEASE #27 and have run into some strange behavior with the SCSI drive. The adaptor is an Adaptec 29160 Ultra160 adaptater, and the drive is a Seagate ST336607LW SCSI-3 device. Problem is that I can rm a 4GB file and afterwards, the free block count won't reflect the erasure. Instead, it displays the same count it showed prior to the 'rm'. The free block count won't be updated until I reboot the system. I checked LINT, and found that the 29160 isn't supported. Only 154x, 274x/284x/294x, and 152x are mentioned there. It's probably a no-brainer at this point, but I want to be sure: could this be the cause of the OS not seeing/showing the updated free block count until after the reboot? Thank you, John -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ | The people and friends that we have lost, the dreams that have faded... | | never forget them. -- Yuna, Final Fantasy X| +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
finding FreeBSD 3.2-RELASE
Hello, I need to create a FreeBSD 3.2 system in order to facilitate the upgrade of one of our important systems, but ftp.freebsd.org doesn't have 3.2-RELEASE anymore. Does anybody know of an FTP server that does still have it? Or of a site that has the distribution ISOs? Any such information would be very much appreciated. Thank you, John -- +---+ | John Fox [EMAIL PROTECTED] |System Administrator | InfoStructure | +---+ | The people and friends that we have lost, the dreams that have faded... | | never forget them. -- Yuna, Final Fantasy X| +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Question about vinum and contents of /dev/vinum
I am copying files over the network -- from a failing drive in production box to a new drive in my workstation, with the intent of swapping these drives. I have been testing transfers with tar, and ran into some problems when it (tar) hit the '/dev/vinum' hierarchy: tar: dev/vinum/plex/virtdisk_plex: minor number too large; not dumped tar: dev/vinum/sd/drive0: minor number too large; not dumped tar: dev/vinum/sd/drive1: minor number too large; not dumped tar: dev/vinum/vol/virtdisk.plex/virtdisk_plex: minor number too large; not dumped tar: dev/vinum/Control: minor number too large; not dumped tar: dev/vinum/control: minor number too large; not dumped tar: dev/vinum/controld: minor number too large; not dumped tar: dev/sa0.ctl: minor number too large; not dumped My question is this: When vinum starts up and can't find some of these device files, will it crash and burn, or will it simply create them? Thanks in advance for any help, John +---+ | I can't believe I'm back on Mars. Three times before this place almost | | killed me. I swore I would never give it another chance to complete | | the job. Humans got no business being here. No business at all. | | -- Michael Garibaldi, _Babylon 5_, The Exercise of Vital Powers | +---+ | John Fox [EMAIL PROTECTED] | System Administrator | Internet Ventures Oregon | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
'origin not recorded' error upon attempt to use pkg_add
Hello! Upon attempts to use 'pkg_add' utility, I get a string of error messages such as the following: pkg_add: package wmtime-1.0b2 has no origin recorded pkg_add: package wterm-6.2.7a1 has no origin recorded It would be only a curiosity were it not for the fact that pkg_add *fails* to add the requested package -- it returns a '1'. So I am very curious about two things: 1) The methodology by which I can correct this error condition. (Most important) 2) Theories about what causes this condition. (Least important) Thank you in advance for any advice, John +---+ | I can't believe I'm back on Mars. Three times before this place almost | | killed me. I swore I would never give it another chance to complete | | the job. Humans got no business being here. No business at all. | | -- Michael Garibaldi, _Babylon 5_, The Exercise of Vital Powers | +---+ | John Fox [EMAIL PROTECTED] | System Administrator | Internet Ventures Oregon | +---+ ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]