tag:blogger.com,1999:blog-14093712.post8240110669900409638..comments2023-08-23T05:08:43.262-07:00Comments on Little Notes: Amanda: simple ZFS backup or S3jmlittle@gmail.com (Joe Little)http://www.blogger.com/profile/09731419203596760536noreply@blogger.comBlogger9125tag:blogger.com,1999:blog-14093712.post-48797039299591516302010-08-06T11:46:47.206-07:002010-08-06T11:46:47.206-07:00Thanks for the post. Regarding the spanning, ther...Thanks for the post. Regarding the spanning, there's work afoot in 3.2 to make this even better-behaved, so that you don't waste space at the end of each tape. I wrote it up <a href="http://code.v.igoro.us/archives/59-Whats-New-in-Amanda-The-End-of-Fragmentation.html" rel="nofollow">here</a>, if you're interested.Unknownhttps://www.blogger.com/profile/07066168408300187392noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-43563420721756922072009-11-24T15:22:38.663-08:002009-11-24T15:22:38.663-08:00auto-tier is what I'm using for D2D backups, b...auto-tier is what I'm using for D2D backups, but it does need to traverse both filesystems to see differences. What you what is auto-sync, a service facility of NexentaStor that just sends those deltas using zfs send/recv or rsync protocols, but based on snapshots. It requires both primary disk and 2nd tier to be NexentaStor-based solutions, and it will mirror the data of the first tier to the second tier using the same snapshots, but the second tier can then have a larger snapshot retention pool.<br /><br />I used auto-tier in my case as our primary storage is still mostly NetApp based. We use NexentaStor for primary storage, but since first tier is hybrid between the two, we need to use auto-tier.jmlittle@gmail.com (Joe Little)https://www.blogger.com/profile/09731419203596760536noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-26013210506783431552009-11-24T14:36:35.967-08:002009-11-24T14:36:35.967-08:00This seems to be my problem. Backing up multi ter...This seems to be my problem. Backing up multi terabytes is taking 4-6 days with a DDT solution just on the DD side. What I really need is a snapshot that only shows the differential (or cumulative) from the last snapshot. Still looking for something more robust.Chuffhttps://www.blogger.com/profile/12989671311273524646noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-87762333991517846452009-07-15T10:49:02.035-07:002009-07-15T10:49:02.035-07:00The snapshots are generated by NexentaStor for you...The snapshots are generated by NexentaStor for you. There are other "services" that can be implemented on standard OpenSolaris, but I use Nexenta to manage the snapshots, retention policies, etc. <br /><br />As for who is handling incrementals, its all in the snapshots. As stated in the article, I only use amanda to go to tape (or S3) for a final archive, so it is always just a full non-incremental backup step. The point to using monthly snapshots is to allow the backup a very long window without worry of data changing. My backups to tape now regularly run weeks at a time for dozens of terabytes.jmlittle@gmail.com (Joe Little)https://www.blogger.com/profile/09731419203596760536noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-44151377701408276412009-07-15T10:39:26.623-07:002009-07-15T10:39:26.623-07:00So do you run a script before amanda kicks off or ...So do you run a script before amanda kicks off or have amanda manipulate the snapshot? <br /><br />Is the snapshot handling the incremental or is amanda? I'm guessing Amanda. <br /><br />This is pretty cool. I never liked the send/recv stuff, that I'm doing in my home spun script.Mostly Nothinghttps://www.blogger.com/profile/07511731036581746798noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-2474258880199477462009-07-15T10:38:09.330-07:002009-07-15T10:38:09.330-07:00So do you run a script before amanda kicks off or ...So do you run a script before amanda kicks off or have amanda manipulate the snapshot? <br /><br />Is the snapshot handling the incremental or is amanda? I'm guessing Amanda. <br /><br />This is pretty cool. I never liked the send/recv stuff, that I'm doing in my home spun script.Mostly Nothinghttps://www.blogger.com/profile/07511731036581746798noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-66090557407345224112009-01-27T09:35:00.000-08:002009-01-27T09:35:00.000-08:00Sorry about Joe! Thanks for clearing that up. I wa...Sorry about Joe! Thanks for clearing that up. I wasn't sure that it would just go through the tapelist. It sounded like it would take a second tape and label it like DailySet1-10-2 if the first one was DailySet1-10. This makes a difference on the number of tapes to use. Again, thanks!Unknownhttps://www.blogger.com/profile/17342786393579824391noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-75058366295164762292009-01-27T09:07:00.000-08:002009-01-27T09:07:00.000-08:00First, I'm Joe :)In amanda you either pre-label th...First, I'm Joe :)<BR/><BR/>In amanda you either pre-label the tapes using "amlabel" and it will span your series, like "CIS-01" to "CIS-40". You define the tape sets and reuse attributes in the amanda.conf file. You can also just fill up a library full of empty tapes and tell newer Amanda versions to auto-label. The key thing to know is that these are amanda file-mark labels at the beginning of the tape. There is separate support for reading and handling the various tape-library bar code labels. I don't bother to try to match that up with Amanda labels other than the names I give them.jmlittle@gmail.com (Joe Little)https://www.blogger.com/profile/09731419203596760536noreply@blogger.comtag:blogger.com,1999:blog-14093712.post-54084701914100404022009-01-27T08:49:00.000-08:002009-01-27T08:49:00.000-08:00Jim, I've not found anything that explains labelin...Jim, I've not found anything that explains labeling for spanning tapes. What did you do for this? Thanks, JimUnknownhttps://www.blogger.com/profile/17342786393579824391noreply@blogger.com