Results 1 to 2 of 2

Thread: Multi-tenant Amanda?

  1. #1
    Join Date
    Sep 2011
    Posts
    2

    Default Multi-tenant Amanda?

    I'm hoping there is an easy answer to this...

    I'm replacing our current backup solution which I use for backing up data for several different customers. I'm moving away from a crusty, failing, tape library to S3 backups instead (got a sweet deal on bandwidth).

    Amanda seems to expect that a "configuration" correspond to a schedule. I can infer this this because when you run "amdump <config>", it runs everything in the disklist under that config. So, basically it looks like I must organize configs by schedule. I haven't found any way to run a partial disklist file, seems to be all or nothing.

    That poses the following issues for me:
    1. I'd like to keep customer data in separate folders in my S3 bucket, for ease of deletion, accounting, etc. If I co-mingle them across configs, I'm not sure if I can keep a single customer limited to a single set of virtual tapes.
    2. When I try to run more than one amdump process, I get a message saying that an Amanda process is already running.

    I have one customer, for instance, where we backup archive logs on an hourly schedule. When the nightly's kick off for all the customers, at all different times according to the customer's scheduled window's for I/O load, those amdump processes for the hourly would either take precedence or fail.

    Basically, this is what I thought I wanted to do, but didn't work right out of the gate:

    /etc/amanda/customer1/amanda.cfg
    /etc/amanda/customer2/amanda.cfg
    /etc/amanda/customer3/amanda.cfg

    This didn't work right away because each customer has several hosts which are all backed up on different schedules. I was thinking of something like this:

    /etc/amanda/hour-00/amanda.cfg
    /etc/amanda/hour-01/amanda.cfg
    ...
    /etc/amanda/hour-23/amanda.cfg

    However, that would be really hard to manage and cause data mingling which I do not want.

    I did find the "inparallel" setting, but it seemed to refer to running parallel backups of all the entries in the disklist under a single config, not across multiple configs.

    Am I asking Amanda to do something it wasn't designed for or going about this the wrong way?

    Bonus question: I need a monthly report by customer data backed up for billing. Best I think I can do it run amreport after each successful backup run, parse the output, push into a DB, and then write a script to query the data out monthly. I didn't find any options for amreport which allowed me to get any data over a time-frame.

    Thanks in advance,
    Dave

  2. #2
    Join Date
    Sep 2011
    Posts
    2

    Default Got most of it...

    In case anyone else runs into this...

    Turns out you can run more than one amdump at a time, but uniqueness is checked by the amdump script based on the log dir:
    ...
    return unless -f $amdump_log_filename || -f $trace_log_filename;
    ...

    So, as long as you set your logdir in amanda.conf to be unique, it will work. logdir "/somepath/customer1"

    Also, you can run a partial disklist. I misinterpreted the man page for amdump. It's pretty simple: amdump <configuration> <host>

    So that just leaves the monthly report. Looks like using Amanda::Report to scrape the logs is probably the best way to go with that. Apparently the enterprise edition has those features built in.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •