Re: Splitting a DLE with regex
Thanks to Gerrit, Brian and Jean-Louis. I'll be responding to them in this message Hello Gerrit, thanks for chiming in!! include ./[a-c].* Maybe you should first try an existing pathname (so no wildcard), just to be sure the ./ is correct here. Good point, thanks! localhost /home/a1 /home { vhost2-user-tar include ./aaronson } 1 local And an amcheck seemed to work: Client check: 1 host checked in 1.220 seconds. 0 problems found. So, it appears that the ./ is correct, which follows what Jean-louis follows with: include must be a glob expression, not a regex, so ./[a-c]* is the correct syntax, Do /home and /home/aaronson are on the same filesystem? df /home df /home/aaronson Yes, they are on the same filesystem: bash-4.1$ df /home /home/aaronson Filesystem 1K-blocks Used Available Use% Mounted on /dev/vda1935102536 420335640 514756896 45% / /dev/vda1935102536 420335640 514756896 45% / Are you using the application 'amgtar' or the program 'GNUTAR'? I *think* that amanda was configured --with-program GNUTAR, but it was compiled before I took over. Is there a way to check, via a debug file or log? Brian added: Remember, to check where you are anchored. I've lost your earlier emails, but you do have to be careful to know what your starting point is. Starting point is /home and then the include directive is for ./[a-c]* And all subdirectories are lower-case, except the numbered which will get ./[0-9]* as their include directive. Thanks so much, ~~Mike
Re: Splitting a DLE with regex
Jean-Louis: In the amandad and selfcheck debug files, do you have amgtar debug files? No amgtar debug files. amadmin CONFIG disklist Performing an amadmin Daily disklist shows that each uses program GNUTAR in it. Checking the amadmin '' version that Gerrit recommended shows: bash-4.1$ amadmin '' version build: VERSION=Amanda-3.3.1 BUILT_DATE=Fri Nov 16 09:09:53 EST 2012 BUILT_MACH= BUILT_REV=4553 BUILT_BRANCH=community_3_3_1 CC=gcc paths: bindir=/usr/local/bin sbindir=/usr/local/sbin libexecdir=/usr/local/libexec amlibexecdir=/usr/local/libexec/amanda mandir=/usr/local/share/man AMANDA_TMPDIR=/tmp/amanda AMANDA_DBGDIR=/tmp/amanda CONFIG_DIR=/usr/local/etc/amanda DEV_PREFIX=/dev/ RDEV_PREFIX=/dev/ DUMP=UNDEF RESTORE=UNDEF VDUMP=UNDEF VRESTORE=UNDEF XFSDUMP=UNDEF XFSRESTORE=UNDEF VXDUMP=UNDEF VXRESTORE=UNDEF SAMBA_CLIENT=UNDEF GNUTAR=/bin/gtar COMPRESS_PATH=/bin/gzip UNCOMPRESS_PATH=/bin/gzip LPRCMD=UNDEF MAILER=UNDEF listed_incr_dir=/usr/local/var/amanda/gnutar-lists defs: DEFAULT_SERVER=vhost2.summersault.com DEFAULT_CONFIG=DailySet1 DEFAULT_TAPE_SERVER=vhost2.summersault.com DEFAULT_TAPE_DEVICE= NEED_STRSTR AMFLOCK_POSIX AMFLOCK_FLOCK AMFLOCK_LOCKF AMFLOCK_LNLOCK SETPGRP_VOID AMANDA_DEBUG_DAYS=4 BSD_SECURITY USE_AMANDAHOSTS CLIENT_LOGIN=amandabackup CHECK_USERID HAVE_GZIP COMPRESS_SUFFIX=.gz COMPRESS_FAST_OPT=--fast COMPRESS_BEST_OPT=--best UNCOMPRESS_OPT=-dc The GNUTAR path is in the paths section. Do the amanda user can read /home? If I put the directories into the disklist, as individual DLEs, like this: /home/aaronson /home/aferton /home/amigos /home/archery /home/blank /home/borders /home/brynwyn /home/caronwyn /home/cirque /home/common Then they are backed up without any issues, but they cause problems with the glob {include ./[a-c]*} However, as the amandabackup user, when I cd /home and then ls -lah I get a permission denied. This could be the reason the glob directive is failing, since it cannot scan the contents of that directory to put together an accurate include list for tar. Unfortunately, each /home/username is owned by username:username as the UID:GID. It now looks like I need to figure out a way to get the amandabackup user access via a groupID. Thanks! ~~Mike
Splitting a DLE with regex
Hello all, Attempting to follow the examples in the amanda documentation and results when I search for it in the mailing list Archives, I encounter problems. I have one DLE (/home) that is approx 200GB and would like to split this out using regex similar to the examples given in: http://archives.zmanda.com/amanda-archives/viewtopic.php?t=6935 http://wiki.zmanda.com/index.php/How_To:Split_DLEs_With_Exclude_Lists#Splitting_it_up.2C_the_easy_part When I create the disklist entry using the hostname, diskname (mnemonic, so the [hostname diskname] pairs are unique) and devicename (using the mountpoint): localhost /home/a-c /home { vhost2-user-tar include ./[a-c]* } 1 local an amcheck returns: ERROR: localhost: [Nothing found to include for disk /home/a-c] There are several subdirectories that would fall within the a-c range: /home/aaronson /home/aferton /home/amigos /home/archery /home/blank /home/borders /home/brynwyn /home/caronwyn /home/cirque /home/common What am I missing? I'd like to split up this monolithic DLE into more manageable portions (a-c, d-f, g-i, and so on...) but can't figure out why amanda is having problems with the regex. Thanks! ~~Mike
Driver exit signals?
Good day! Noticed today that our amanda (3.3.1 on CentOS) complained in the daily report: === driver: Taper protocol error driver: going into degraded mode because of taper component error. === This happened after setting up an exclusion list, so now I'm trying to figure out why creating that has caused the issue (or if it's even related at all). :) Looking through the amdump logfile for this run, I see: === WARNING driver taper pid 16514 exited with signal 6 === Searching, I can't seem to find a list of the exit signals and their meanings. I thought I saw them listed on the wiki site in the past, but can't seem to find them now. Are they still there? Thanks, ~~Mike
Re: Adding/removing Disks?
On 05/09/2012 08:01 AM, Greg Copeland wrote: I have a need to add a new disk and remove an existing disk. Can this be safely done by simply adding and removing the definitions in the configuration? If you're simply adding/removing disks to be backed-up, you'll want to edit the file that contains your DLE's. After removing/adding the appropriate entries (don't forget the appropriate dump type) be sure to run an amcheck afterwards to verify your changes. If all goes well, amanda will create the index files and perform a level 0 dump of the new disks on the next run for that configuration. If you want to add/remove holding disks you'll need, if the mountpoints will be different, to edit your configuration files to change the information about the holding disks (including reserve amount, mountpoint, etc). I would recommend checking to make sure that there isn't any data in the holding disk first, by doing an amflush. Good luck! ~~Mike
Re: 'Bogus-Month 0,0' reports for a job that doesn't exist?
Just a follow-up... Thank you for the suggestions, Jean-Louis. I removed the log file, ran amcleanup, and made sure that all amanda processes on the server were stopped. After that, the Daily backup job was completed successfully again, and have been working correctly since. Thank you, ~~Mike
Re: 'Bogus-Month 0,0' reports for a job that doesn't exist?
On 01/13/2012 10:39 AM, Jean-Louis Martineau wrote: It is because the correct timestamp line is not include in the log file. Any others error in the report? Kill all amanda processes run amcleanup Remove the 'log' file from the logdir if you have one. Thank you, Jean-Louis! That information will allow me to dig deeper and try to find out why this is occurring. I will follow your suggestions! Yes, there are estimate time-out errors in the report, such as: = FAILURE AND STRANGE DUMP SUMMARY: planner: ERROR liver2NAK : amandad busy planner: ERROR patchNAK : amandad busy planner: ERROR jazzNAK : amandad busy planner: ERROR vector1NAK : amandad busy planner: ERROR Estimate timeout from hedwig hedwig /usr lev 1 FAILED [disk /usr, all estimate timed out] hedwig /var lev 1 FAILED [disk /var, all estimate timed out] hedwig / lev 0 FAILED [disk /, all estimate timed out] planner: ERROR Estimate timeout from hedwig vector1 /Summer lev 0 FAILED [Estimate timeout from vector1] vector1 /Users/becky/Public lev 1 FAILED [Estimate timeout from vector1] = and so on. This is from amanda community server 2.4.5 on a FreeBSD 6.1 system. Thanks, ~~Mike
'Bogus-Month 0,0' reports for a job that doesn't exist?
Good morning all! I've started getting these reports in the morning, within the past week, but I can't seem to figure out why. It seems that a dump is happening overnight but I can't figure out where it's coming from. The only nightly amanda job in my crontab is the Daily, which kicks off at 03:30, but goes into degraded holdingdisk mode since the Bogus-Month job is accessing the tapedrive. I'm concerned that my Weekly, on Sunday morning, is going to be interrupted as well. I've done mailing list archive searches to no avail. I see references from years ago, but no indication what's causing them, and this is a new occurrence for me. http://archives.zmanda.com/amanda-archives/viewtopic.php?t=3243sid=e3f27757b4ea2a11fd205168a826ad48 I've searched the wiki.zmanda.com site for bogus without any results. The root crontab and the amanda user (operator)'s crontab are both logical and no Bogus-Month run is listed. I don't have a Bogus-Month config directory, either, which is even more confusing. Can someone please help shed some light on this for me? Thanks! ~~Mike
3.3.0 on Mac OS X (Lion) conflicting types for 'semaphore_t' error
Good morning! New install of amanda client on a brand new computer (still picking pieces of styrofoam packing off the keyboard). Installed Xcode 4.1 this morning, then gathered the required dependencies for amanda: gettext (0.18.1.1) glib (2.29.2) libiconv (1.13.1) pkg-config (0.25) amanda (3.3.0 src) After installing the dependencies, I configured the amanda client: ./configure --with-user=amanda --with-group=wheel --prefix=/usr/local --without-server --with-config='Desktop' --with-gnutar=/usr/bin/gnutar --with-index-server=jazz.summersault.com There were no errors. Upon performing a make, I receive: = libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../config -I../../common-src -I../../common-src -I../../xfer-src -I../../gnulib -I../../ndmp-src -I/System/Library/Perl/5.12/darwin-thread-multi-2level/CORE -fno-strict-aliasing -D_GNU_SOURCE -I/usr/local/include -D_REENTRANT -I/usr/local/include/glib-2.0 -I/usr/local/lib/glib-2.0/include -DSWIG -g -O2 -fno-strict-aliasing -MT xferwrap.lo -MD -MP -MF .deps/xferwrap.Tpo -c xferwrap.c -fno-common -DPIC -o .libs/xferwrap.o In file included from ../../xfer-src/element-glue.h:28, from ../../xfer-src/amxfer.h:28, from xferwrap.c:24: ../../common-src/semaphore.h:33: error: conflicting types for ‘semaphore_t’ /usr/include/mach/mach_types.h:126: error: previous declaration of ‘semaphore_t’ was here make[5]: *** [xferwrap.lo] Error 1 make[4]: *** [all] Error 2 make[3]: *** [all-recursive] Error 1 make[2]: *** [all] Error 2 make[1]: *** [all-recursive] Error 1 make: *** [all] Error 2 bash-3.2# = Doing a search for the semaphore error message, I see this thread: http://forums.zmanda.com/showthread.php?t=3701 I downloaded the patch file, did a make distclean to clean up. I applied the patch, and re-ran the configure script (same arguments as above) without errors. When I make again, I get: = libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../config -I../gnulib -fno-strict-aliasing -D_GNU_SOURCE -I/usr/local/include -D_REENTRANT -I/usr/local/include/glib-2.0 -I/usr/local/lib/glib-2.0/include -Wall -Wextra -Wparentheses -Wdeclaration-after-statement -Wmissing-prototypes -Wstrict-prototypes -Wmissing-declarations -Wformat -Wformat-security -Wsign-compare -Wfloat-equal -Wold-style-definition -Wno-strict-aliasing -Wno-unknown-pragmas -g -O2 -fno-strict-aliasing -MT protocol.lo -MD -MP -MF .deps/protocol.Tpo -c protocol.c -o protocol.o /dev/null 21 mv -f .deps/protocol.Tpo .deps/protocol.Plo make[3]: *** No rule to make target `semaphore.c', needed by `semaphore.lo'. Stop. make[2]: *** [all] Error 2 make[1]: *** [all-recursive] Error 1 make: *** [all] Error 2 = I see that there is instructions to edit the Makefile.in file and replace semaphore with amsemaphore in appropriate places. However, I don't know what would be appropriate places to make the replacement. Can someone provide some insight? I see that Jean-Louis stated he committed a patch for this, but unfortunately, it seems to not have gotten included in the 3.3.0 that I downloaded this morning. Thank you! ~~Mike
includefile functionality within disklist.conf?
Good morning! Does the include function, which works within the amanda.conf file to allow external files to be included in the configuration, work within the disklist file? This would allow easy editing of the DLEs for a specific client system without my counterparts mucking through multiple clients' DLEs. I have tried, using the 'includefile path/to/file.conf' syntax, but get an error on the amanda 2.4.5 server while performing an amcheck. Does this funtionality work with newer versions? I can't seem to find an answer on the amanda wiki (of course, I may not be looking in the right place!). Thanks! ~~Mike
Perl error while compiling amanda
Greetings, all. Attempting to perform a make of amanda-client on a FreeBSD 8.1 system, to update from 2.5.1p2, I receive the following error: checking for perl version greater than or equal to 5.6.0... Segmentation fault (core dumped) However, a real quick (perl -v) check reveals that we're at perl 5.8.8 on that system. Hrmm. The error message mentions to email kuriyama@freebsd about the issue with logs and such, which I plan on doing, but I wanted to check here to see if others had run into this before, and if there's a known error or work-around for it? And now, off to do some googling and see if there's reports of this behaviour. :) Thanks! ~~Mike
Re: tar 1.2.6
On 03/22/2011 02:23 AM, gene heskett wrote: So finally, linux once again has a tar that amanda can use. Which version of amanda are you using?
Error: disk /usr, all estimate failed
Good afternoon! First post to the group, and am VERY new to Amanda, so please be gentle. :) I've inherited an Amanda setup that includes a central server (FreeBSD 4.10) running Amanda 2.6.1p1, and several clients running Amandad with the same version number, all writing to a DST tape drive via SCSI. Everything has been working fine, until recently (hence my post here), when the server was upgraded to FreeBSD 8.1, and an external HDD attached at /home3/ for the Amanda holdingdisks. Everything seems to be working fine (surprising, given the large jump in OS, yes) with regular incremental and even our weekly and Monthly runs going off successfully. But now one of my DLE's for the Amanda server itself shows an error (/usr lev 0 FAILED [disk /usr, all estimate failed]) in the daily report, even though the selfcheck.*.debug file shows no errors, and an amcheck command shows no problems. We are using tar to perform the dump, since it obeys an exclude file listing for certain directories. What information can I offer to assist you with assisting me? Again, I'm a newbie, so please forgive if I missed something obvious. Thanks! ~~Mike
Re: Error: disk /usr, all estimate failed
Okay, looking further, I see the following in the file server:/tmp/amanda/sendsize.20110207020002.debug - sendsize[77723]: time 5.145: getting size via gnutar for /usr level 0 sendsize[77723]: time 5.311: spawning /usr/local/libexec/runtar in pipeline sendsize[77723]: argument list: /usr/bin/tar --create --file /dev/null --directory /usr --one-file-system --listed-incremental /usr/local/var/amanda/gnutar-lists/server_usr_0.new --sparse --ignore-failed-read --totals --exclude-from /tmp/amanda/sendsize._usr.20110207020007.exclude . sendsize[77723]: time 5.464: tar: Option --listed-incremental is not supported sendsize[77723]: time 5.464: Usage: sendsize[77723]: time 5.465: List:tar -tf archive-filename sendsize[77723]: time 5.465: Extract: tar -xf archive-filename sendsize[77723]: time 5.465: Create: tar -cf archive-filename [filenames...] sendsize[77723]: time 5.465: Help:tar --help sendsize[77723]: time 5.466: . sendsize[77723]: estimate time for /usr level 0: 0.154 sendsize[77723]: no size line match in /usr/bin/tar output for /usr sendsize[77723]: . sendsize[77723]: estimate size for /usr level 0: -1 KB Just a guess here, but I presume that since amanda is waiting on a response from tar, and tar has errored out, that amanda skips this item in it's DLE because of this error. Does this sound reasonable? Now, if that's the case, I either need to figure out how to edit what commands amanda is sending to the tar program and remove the offending commandline flag, or downgrade the version of tar that is on the newly updated server. Thoughts? Thanks, ~~Mike