Re: Patch for bugs in json.pm

Hi Jean-Louis,

I came up with a combination of your first and third examples. I like the
labels on the first example; it's consistent with the rest of the file and
it provides a simple way to loop on any of the keys. I also like the idea
of combining strange and failed into the same data structure. So here's
some real output (the hashes return keys in random order):


   "dump_details" : [
      {
         "strange" : [
            "sendbackup: start [aluminum.abc.example.com:/u00 level 1]",
            "sendbackup: info BACKUP=/bin/gtar",
            "sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -xpGf -
...",
            "sendbackup: info COMPRESS_SUFFIX=.gz",
            "sendbackup: info end",
            "? /bin/gtar: ./oradata/sbmmgr/sbmmgr_srl0.f: file changed as
we read it",
            "? /bin/gtar: ./oradata/sbmprd/sbmprd_srl0.f: file changed as
we read it",
            "| Total bytes written: 1161318400 (1.1GiB, 50MiB/s)",
            "sendbackup: size 1134100",
            "sendbackup: end"
         ],
         "disk_name" : "/u00",
         "backup_level" : "1",
         "hostname" : "aluminum.abc.example.com"
      },
      {
         "disk_name" : "\"c:/\"",
         "failure" : [
            "sendbackup: start [neon.abc.example.com:\"c:/\" level 0]",
            "sendbackup: info BACKUP=pkzip",
            "sendbackup: info RECOVER_CMD=Extract with zmanda windows
client or unzip program",
            "sendbackup: info end"
         ],
         "backup_level" : "0",
         "error" : "[data timeout]",
         "hostname" : "neon.abc.example.com"
      },
      ...

(I don't have an example of both a failure and a strange on the same DLE)


If you're good with this arrangement, I'll submit the diff.

Thanks,

On Wed, Mar 2, 2016 at 6:56 AM, Jean-Louis Martineau <
[email protected]> wrote:

> Michael,
>
> I will accept your change if you send a patch, read at the end of the
> email for more strange_details improvement.
>
> Jean-Louis
>
> On 01/03/16 06:17 PM, Michael Burk wrote:
>
> Hello,
>
> I appreciate the JSON report format option, as it gives me a flexible way
> to do things like turn the data into an HTML report. I have noticed a few
> oddities in the JSON output, and have a couple other suggestions to improve
> it. I don't know if you consider the JSON format to be frozen yet or not.
> If you are interested in these changes, I'd be happy to submit a diff.
>
>
> ## Oddities
>
> - section labeled "statistic" instead of "statistics"
>   (unexpected, inconsistent with other section keys)
>
> - some numerical statistics have quoted values (inconsistent):
>     - run_time
>     - tape_used.incr,full,total
>     - estimate_time
>     - avg_compression.incr,full,total
>     - avg_dump_rate.incr,full,total
>
> - "Avg_tape_write_speed" is the only capitalized key (inconsistent)
>
>
> ## Improvements
>
> - Failed/Strange Dump Details are copies of these text sections in the
> plain
>   text version of the report (arrays of text). It would be better to make
> them
>   arrays of arrays of text. In other words, each DLE output would be an
> array
>   element, like so:
>
>    "strange_details" : [
>         [
>           "  aluminum.abc.example.com /u00 lev 1 STRANGE",
>           "  sendbackup: start [aluminum.abc.example.com:/u00 level 1]",
>           "  sendbackup: info BACKUP=/bin/gtar",
>           "  sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -xpGf -
> ...",
>           "  sendbackup: info COMPRESS_SUFFIX=.gz",
>           "  sendbackup: info end",
>           "  ? /bin/gtar: ./oradata/sbmmgr/sbmmgr_srl0.f: file changed as
> we read it",
>           "  ? /bin/gtar: ./oradata/sbmprd/sbmprd_srl0.f: file changed as
> we read it",
>           "  | Total bytes written: 1161318400 (1.1GiB, 50MiB/s)",
>           "  sendbackup: size 1134100",
>           "  sendbackup: end",
>         ],
>         [
>           "  boron.abc.example.com /u00 lev 1 STRANGE",
>           "  sendbackup: start [boron.abc.example.com:/u00 level 1]",
>           "  sendbackup: info BACKUP=/bin/gtar",
>           "  sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -xpGf -
> ...",
>           "  sendbackup: info COMPRESS_SUFFIX=.gz",
>           "  sendbackup: info end",
>           "  ? /bin/gtar: ./app/oracle/product/
> 11.2.0.4/network/log/listener.log: file changed as we read it",
>           "  | Total bytes written: 1739325440 (1.7GiB, 30MiB/s)",
>           "  sendbackup: size 1698560",
>           "  sendbackup: end",
>         ],
>             ...
>     ]
>
>
>
> How about the following for strange_details?
>
> "strange_details" : [
>  {
>     "host": "aluminum.abc.example.com",
>    "dle" : "/u00",
>    "level" : 1,
>    "strange" :
>     [
>           "  sendbackup: start [aluminum.abc.example.com:/u00 level 1]",
>           "  sendbackup: info BACKUP=/bin/gtar",
>           "  sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -xpGf -
> ...",
>           "  sendbackup: info COMPRESS_SUFFIX=.gz",
>           "  sendbackup: info end",
>           "  ? /bin/gtar: ./oradata/sbmmgr/sbmmgr_srl0.f: file changed as
> we read it",
>           "  ? /bin/gtar: ./oradata/sbmprd/sbmprd_srl0.f: file changed as
> we read it",
>           "  | Total bytes written: 1161318400 (1.1GiB, 50MiB/s)",
>           "  sendbackup: size 1134100",
>           "  sendbackup: end"
>     ]
>   },
>   { ...
>   }
> ]
>
> or the following where host, dle and level values are keys of hashes:
> "strange_details": {
>   "aluminum.abc.example.com": {
>     "/u00" : {
>       1: [
>           "  sendbackup: start [aluminum.abc.example.com:/u00 level 1]",
>           "  sendbackup: info BACKUP=/bin/gtar",
>           ....
>           ]
>      }
>   }
> }
>
>
> Then why two keys at the top level for strange/failed, would it be better
> to group them like the following:
> "details": {
>   "aluminum.abc.example.com": {
>     "/u00" : {
>       1: {
>          "strange" : [
>             "  sendbackup: start [aluminum.abc.example.com:/u00 level 1]",
>             "  sendbackup: info BACKUP=/bin/gtar",
>             ....
>          ],
>          "failed": [
>             ...
>          ]
>      }
>   }
> }
>