| # |
Item |
References |
Comments from others |
DMF - Management |
| 33. |
Being able to see and alter the DMF request queue. This should
include an interface suitable for use in a shell script.
|
2011 #2
|
|
| 34. |
Add a site-specified comment field to the tape entries in the
dmvoladm database, of at least 8 characters.
|
|
|
| 35. |
Being able to override HLOCK behaviour with a dmf.conf setting,
to prevent HLOCK being cleared on VSNs when DMF is restarted.
|
|
|
| 36. |
Allow the dmmove and dmget commands to specify the preferred VG
to use for recalls.
|
2011 #10
|
Implemented at ISSP 3.0
|
| 37. |
Quotas for DMF managed filesystems taking into account lower tier
storage space (tapes). Ideally both reporting and enforcement
of hard/soft limits. Possibly integrate into xfs_quota.
|
2011 #5
|
|
| 38. |
Up/down scrolling in command history in dm*adm utilities using
readline or similar.
|
2011 #6
|
Bill Ryder: the rlwrap wrapper works with the dm*adm
utilities - it's in the opensuse repository and maybe others.
|
| 39. |
(deleted)
|
|
|
| 40. |
I'd like better control over the way dmmove does and does
not do the internal equivalent of a dmdidle, to help in the
transition from one tape type to another. I have encountered
problems in the past in trying to keep a constant stream of data
moving especially with data migrations from small tapes to large
tapes. I ended up with an hourly cron job that ran dmdidle on the
destination VGs to prevent huge delays in the dmmove completing.
|
|
|
| 41. |
Finding new releases inside SupportFolio (even the revamped
version) is hit and miss; you just wander around until you
accidentally find what you want. Then you have to download it to
a workstation, transfer it to the DMF server and install with RPM,
avoiding the architecture which doesn't apply to your system.
Why not use a repository that can be used with YAST Online Update (YOU)?
|
|
Implemented at ISSP 3.0
|
| 42. |
The packaging of patches is overly complicated, especially for
DMF clients with the Linux ones being the worst. A tarball
containing client RPMs to be installed on the server(!) which
install further tarballs containing RPMs for installation on
the clients.
|
|
|
| 43. |
DUMP_MIGRATE_FIRST does not consider policies when migrating files
(namely what files to migrate), make it evaluate them.
|
|
|
| 44. |
"dmput -r" releases all data blocks, making the file OFL. But a
release performed by dmfsfree may leave a file in PAR state,
if the site policies specify this. It would be good to have a
new parameter to dmput to allow the user change a file from DUL
to PAR according to these site policies.
|
|
|
DMF - Performance |
| 45. |
If a recall is received requiring a tape currently in use for
migrations, the migrations should switch to a different tape to
avoid delaying the recall. This is becoming more important as
tape capacities grow over time.
|
2011 #18
|
Implemented at ISSP 3.0
|
DMF - End-User |
| 46. |
DMF-aware Apache capable of displaying a "please wait" message
while an offline web page is being recalled, and then replacing
it with the desired page. Possibly better, issue error codes 300,
429 or 102. And maybe an "offline" indicator on auto-index pages.
|
|
|
OpenVault & Tapes |
| 47. |
My understanding is that DMF can't use dual robots in a T-finity
library without partitioning the library? We have 28 drives and
a very random read workload and we think dual robots might be a
good idea.
|
|
|
Backups |
| 48. |
More flexible way of selecting what not to backup, currently
there is only one obscure way to tell the backup script which
files not to backup (file flags and xfsdump -e).
|
|
|
| 49. |
Implementing a database for backups with appropriate utilities
where I could list the contents (like what files are in the dump)
and details of specific backups without having to mount the tape.
|
|
|
| 50. |
Ability to backup only inodes even when the files are in REG state.
|
|
|
| 51. |
Currently xfsdump uses at least one new tape for backups per task
group, regardless of dump size. Make it intelligently manage the
pool of backup tapes.
|
|
|
| 52. |
Better logging of backup operations - persistent logs and ability
to select the log level.
|
|
|
| 53. |
One of our libraries went offline for a few days. After 2 days
we managed to accumulate 40 TB of data that had a single copy
written out (the 2nd copy could not be written out as the library
was offline). The problem is that the xfsdumps will ignore all
the files that are not in OFL or DUL state and since the 40TB in
question will remain in MIG sate until the 2nd copy is written
out, those inodes are not backed up. The inodes (not the data)
should be included in the dump file.
|
|
|
| 54. |
Allow non-DMF use of the VOL database, to manage tapes used
for backups.
|
|
|
Documentation |
| 55. |
I would like to see more info on Administrator processes/activities
once DMF is installed, configured and running, i.e. what things
need to be done on a daily, weekly, monthly basis.
|
|
|
| 56. |
Also a good (detailed) guide to troubleshooting DMF
problems/issues.
|
|
|
