With the 2.5.2_p1-r3 version all works but after the upgrade to 2.5.2_p1-r4 amanda stops to work returning "Device or resource busy" at every tape access (amcheck, amdump, amflush) Fortunately a simple downgrade restore the normal functionalities. Please ask me for info that could be useful to fix this bug
I've been using this for a while with no such trouble: Tue Jul 1 02:48:41 2008 >>> app-backup/amanda-2.5.2_p1-r4 amanda@archivaris ~ $ myamcheck Amanda Tape Server Host Check ----------------------------- Holding disk /mnt/dump/amanda: 107269044 kB disk space available, using 52428800 kB as requested read label `ORXNET08', date `20080711' NOTE: skipping tape-writable test Tape ORXNET08 label ok Server check took 0.013 seconds Amanda Backup Client Hosts Check -------------------------------- Client check: 5 hosts checked in 0.626 seconds, 0 problems found (brought to you by Amanda 2.5.2p1) Please post your `emerge --info' and describe the tape hardware you use (including the tapetype and tapedev settings from your amanda.conf).
Are you using xinetd or daemontools to start amanda daemons? I'm actually using daemontools and nc, because in the past I have found a env-related problem with the run script suggested in the amanda documentation I think my actual issue could be related with daemontools/nc too.
You said if you downgrade the problem goes away, downgrade to what? I use xinetd for my various amanda deployments (2xLTO1,2xLTO3).
(In reply to comment #3) > You said if you downgrade the problem goes away, downgrade to what? Downgrade to 2.5.2_p1-r3, as I told in the first comment this version works well (naturally I haven't made any change in the configuration)
Ok, could you please try -r4, but comment out line 224 (--with-maxtapeblocksize) or the chg-multi fix (test individually). If the maxtapeblocksize resolves it, next try different values instead of commenting out. Could you please provide us with more information about your setup as well? Tape hardware, spool, emerge --info.
Please do what i suggested in my last comment and report back, otherwise i'll close this bug as NEEDINFO.
(In reply to comment #6) > Please do what i suggested in my last comment and report back, otherwise i'll > close this bug as NEEDINFO. I'm spending my holidays, I'll try it as soon as possible... in September ;)
It's mid september now, and no response. Please test on 2.6.0p2
(In reply to comment #8) > It's mid september now, and no response. > Please test on 2.6.0p2 Using 2.6.0p2-r1 I obtain: Error reading 524288 bytes from /dev/nst0: Device or resource busy Error reading Amanda header. Should I try with your suggestions on comment #5?
Commenting out "--with-maxtapeblocksize" row in the ebuild solves the problem. Which value should I try with? PS: sorry for the delay for the info
For values, please test powers of two from 16k to 512k. Also please specify what blocksize you have configured for your tapes in the amanda tapetype.
I've added AMANDA_MAX_TAPE_BLOCK_KB=32 in my make.conf to make the existing tape-set work
Ok, that's the make.conf, what other values there worked or didn't? Also, please paste your tapetype definition from amanda.conf, it's critical to debug this. Here's the one I use: define tapetype ULTRIUM-TD3 { comment "IBM ULTRIUM-TD3 with LTO3 media, hwcompression off" length 402432 mbytes filemark 0 kbytes speed 69766 kps blocksize 32kb readblocksize 32kb }
No response from user.
