Manual expiration of savesets via gui?

There no way to expire certain save sets from the GUI.

However, the fastest way is simply to relabel the volume.

  !!! This of course only makes sense if no other valuable save sets are stored on the same volume. !!!

For such case, simply use the command line. This is usually a 2 step process:

1. Check the ssid (the cloneid will ensure that a potential duplicate on tape will not be deleted as well)

   mminfo -q "family=disk, " -r "ssid,cloneid" -xc/ [> file]

2. Delete the save set

   nsrmm [-y] -d -S ssid[/cloneid]

Of course you can script that.

For more details please read the Command Line Reference.

Hello bingo. I did ultimately end up using command line, but not quite with the results I was seeking, so I may have missed a step. I should not be comparing Netbackup with Networker, but it's where I'm coming from, so it's a good tool. In Netbackup, you can query the client or dataset using multiple variables, right-click on the images you want to remove, and select "expire now." Then, you can refresh the DataDomain intereface, and see that the size was instantly reclaimed as "cleanable" space. Not "available", but "cleanable".  That actual space will not become "available" until the next cleaning runs against the DataDomain.

So, on the Networker server, I ran  ~>nsrmm -d -S 1707395804 against that ssid, it asked me "Delete file and media index entries for save set `1707395804'?", I confirmed, and it did become no longer available when I ran another search for it in Networker. I did the same thing against the clone ID's. However, the "cleanable" space showing in my DataDomin interface did not change. The saveset was approx 14tb in size, so I feel as though the metadata was removed from Networker, but the space is still used on the DataDomain. Do you think I missed a step?

There are several 'cleaning' processes which run automatically but not right away.

For NW, you could also start the "Index Manager" (nsrim) manually. This should reclaim the space on the NW disk volume.

The default cleaning cycle for the DD is setup for tuesday - I do not know any way to start it from Networker. I think this is a good idea in general as it will not cause contention with other backup/restore jobs.

Of course you can also manually start this process on the DD admin GUI at any time.

Todd,

It’ shard when you’re not familiar with a product and within NetWorker (and I am sure NetBackup also) there are several ways to achieve the same result – determining which best suits you is the trick!

If you have already deleted the saveset manually then NetWorker has no way of letting the Database know during its housekeeping that the saveset is gone and you have overridden the process which should have let the Data Domain know it can release the space from the saveset. It’s best to set is to recyclable and then the housekeeping process which mark savesets for expiry/deletion will take effect and tell the Data Domain the file can be released for cleaning.

You can recover from this situation, I would recommend running the following command: - nsrstage –C -V

Volume name is where the savesets were (hopefully you know this).

This should compare the file system entries and the Media Database entries on DD/NW and sync back up between the hardware/software.

Regards,

Bill Mason

I may be repeating what has already been said? This is the process I use.

Manual Cleanup.

(Please note auto cleanup runs every Tuesday.)

In the GUI you can select a volume, look at the savesets that are older and can be purged. Please note the Saveset ID, and see that the status is browseable. Then, you can run the following command:

nsrmm  -o recyclable -S SSID

This should make the saveset recyclable. Repeat the process for other savesets that can be recycled.

You can run nsrim -X

after recycling the savesets. That should show as cleanable space immediately. That should save you 24 hrs..

Filesys show space        

It should show how much data can be cleaned under the Cleanable GB Column.

Filesys clean start

To start the cleaning process.

Filesys clean watch

To watch the cleaning progress.

Repeat as necessary.

William,

If I understand you, then what I had already done was remove the savesets, but Networker doesnt know that I have, and will not tell DataDomain to make that space reusable? I ran the nsrmm -d -S command, pointing at a txt file containing many ssid's, and ran that through a batch file to delete them all.

If I have done this wrong, then I need to run nsrstage –C -V PDCVMFS01 (<-volume name), to resync whats in the datadomain with what is in the Networker index. Correct? If so, I cannot get that command to work. I may be missing some switches here. Netbackup spoiled me, and I'm not much of a command line dude....yet.

C:\>nsrstage –C -V PDCVMFS01

39094:nsrstage: must specify one of -m or -C.

usage: nsrstage [-v] [-F] [-d] [-s server] [-J storage-node] [-b pool] [-y retention time] -m -S [-o] {[-I] -f file | ssid...}

   or  nsrstage [-v] [-s server] [-J storage-node] [-y retention time] -C -V volume

ttobey,

Can the nsrmm  -o recyclable -S SSID command be used with a text file of ssid's to make many recyclable at once? When I need to expire images manually, its never going to be one....it will be thousands.

Thanks again, all!

Todd

- I doubt you need to run post cleanup as nsrim runs will take care of that in subsequent runs

- use nsrclone -P -V

- you can use -o recyclable too - but you can also use what you used so far

Todd,

I will try and explain some terms to you first to see if this helps your understanding: -

Client File Index (CFI) has the details if a specific clients saveset backups (all files included in a particular backup including timestamps, directory path etc). Until a saveset passes its browse period the details of the files etc are retained then removed once passed from the CFI. We still have the data until its passed its retention period but no longer have all the file details so without manual action to recover CFI or knowing the exact path required we would recover the whole saveset as the detail is not there.

Media Database (MDB) has all the details of the backups for all the savesets from client backups and which volumes they are associated with. We don’t store individual file details here but do have different information regarding the no of files, size, various timestamps (backup, browse, retention, clone times) etc.

Nsrim process runs daily and looks for savesets which are passed their browse period also those which are passed their retention period.

Ones passed browse have the CFI entries removed and their status changed from browsable to recoverable, ones passed retention with no dependancy’s are marked as eligible for recycle. On Advanced File System/file devices and DDBoost devices this the trigger to release the space on the device and allow the disk space to be freed up and the MDB entry is removed also (on DDBoost this works slightly different as the Data Domain cleaning has to run but the cleanable space value should show what is possible to be released on the Data domain itself). For tape volumes – all savesets on a tape must be recyclable before it is eligible to be re-used. When a tape is relabeled all the savesets on that volume are removed from the MDB and the volume information is updated (new volid, label date etc – it appears like a delete and create).

So in your case when you ran the script you deleted the Media Database entries manually so they now no longer exist and therefore cannot be processed by nsrim. This is why you need to do the extra step of running the nststage command as this compares what the DD has for the volume and then if there is no entry in the MDB within NetWorker tells the DD that the savesets are no longer required. This will free up the space.

I am not sure why the command is not working if you volume name is specified correctly – are you running this from the NetWorker server? If so it should work without any extra flags.

If you have ram the nsrmm command to mark the savesets as recyclable rather than doing a delete no further actions should have been necessary as the next automatic run of nsrim would have processed all these and informed he Data Domain we no longer required.

If you cannot get the command to run I would suggest opening a call with EMC Support – it should be quick to resolve.

Regards,

Bill

William Mason wrote: So in your case when you ran the script you deleted the Media Database entries manually so they now no longer exist and therefore cannot be processed by nsrim.  This is why you need to do the extra step of running the  nststage command as this compares what the DD has for the volume and then if there is no entry in the MDB within NetWorker tells the DD that the savesets are no longer required.

BIll, I have question :-D  Isn't that nsrim will run in phases where after checked retention and dependencies and make changes to flags, will trigger cleanup of devices (so essentially the same things as nsrclone -P -W or nsrstage -C -V ).  If so, when he runs nsrim -X (or when NW runs it itself - which happened many times since this thread started) he should also get devices cleaned up.

As for the error you got with -C -V... I noticed that "-" is different lenght when you mentioned here so I suspect you did copy and paste or something else which might have used character which looks the same, but it is not the same and therefore nsrstage command doesn't like it (or recognize it).

Ok, rolling eyes at myself!

E:\Program Files\EMC NetWorker\nsr>nsrstage -C -V PDCVMFS01.001

E:\Program Files\EMC NetWorker\nsr>

I retyped everything in, and it appears to have run this time. Took about a minute. Just for my understanding, this should "sync" (for lack of proper terminology), what is on the DD with the Networker client index, and allow the datadomain to clean images from the savesets that i manually expired?

H and Todd,

I have checked and H is correct and my last statement was not 100% accurate – Disk volumes will be automatically cleaned as part of the recover space checks the volume contents against the MDB not the other way around. However space recovery can be affected by: -

1. A read process on the volume (clone/stage, recover), this does not allow saveset to be removed in case it affects the list of savesets which these processes need to take action on.

2. If the volume is marked as “Scan Required”

So this could be a possible cause of why the space is not released. You can check with mminfo –q scan=true –r volume,pool command. This should list volumes which have the scan required flag set.

Thanks also for spotting the possible syntax from copy/paste ☺

Regards,

Bill

Hi Todd,

Swap Client index for Media Database and you are correct. ☺

Did it work? Do you now have more cleanable free space on the DD for when cleaning job next runs?

Regards,

Bill Mason

Hi all,

Sorry for the late follow-up on this. I have been having to maintain free space on the DataDomain by running cleaning on the Symantec Netbackup side, as it will make expired data 'available' in real time on the datadomain. Im still not convinced that I am doing this correctly with Networker.

Here's what i do.

I check and record the size of the Datadomain, both available size AND cleanable size.

Let's say 7.64tb available and 2.23 cleanable

I generate a list of the SSID's I need to expire (around 3000 savesets, so a significant amount) and save it as ssid.txt.

I then created a bat file containing this string -  For /F %%a in (ssid.txt) do nsrmm -d -y -S %%a

When I run that bat file from the same location as the ssid.txt file, I see it scrolling through each ssid with these messages. (It takes a couple hours).

C:\>ssid_exp.bat

C:\>For /F %a in (ssid.txt) do nsrmm -d -y -S %a

pdc00nwkv801.ohlogistics.com:vm:501940a8-843f-d18f-29e6-62111a6021ff:pdc00vmca802w.ohlogistics.com, 0 browsable cycle(s), 0 browsable files of 540 total, 2831 GB recoverable of 2831 GB total

pdc00nwkv801.ohlogistics.com:vm:501940ca-bafb-3f6d-0ed5-69764c17649b:pdc00vmca802w.ohlogistics.com, 0 browsable cycle(s), 0 browsable files of 540 total, 1698 GB recoverable of 1698 GB total

pdc00nwkv801.ohlogistics.com:vm:5019414b-3f55-7ed6-20a8-aaa5897f68b7:pdc00vmca802w.ohlogistics.com, 0 browsable cycle(s), 0 browsable files of 160 total, 503 GB recoverable of 503 GB total

When these all complete, I then run...

C:\>nsrim -X

pdc00nwkv801.ohlogistics.com:cp.20150728130127, 0 browsable cycle(s), ad hocs, 0 browsable files of 183 total, 2821 MB recoverable of 2821 MB total

pdc00nwkv801.ohlogistics.com:cp.20150729140125, 0 browsable cycle(s), ad hocs, 0 browsable files of 190 total, 3026 MB recoverable of 3026 MB total

pdc00nwkv801.ohlogistics.com:cp.20150730130133, 0 browsable cycle(s), ad hocs, 0 browsable files of 194 total, 3171 MB recoverable of 3171 MB total

pdc00nwkv801.ohlogistics.com:cp.20150731130145, 0 browsable cycle(s), ad hocs, 0 browsable files of 197 total, 3243 MB recoverable of 3243 MB total

pdc00nwkv801.ohlogistics.com:cp.20150801140213, 0 browsable cycle(s), ad hocs, 0 browsable files of 197 total, 3268 MB recoverable of 3268 MB total

.......and it runs for awhile, and then completes.

Then, when I look at my DataDomain, it shows 7.64tb available and 2.23 cleanable....IDENTICAL to what it was before I started. This is why I think something isnt right.

When working with emc before, they said my bat file should have this line in it instead.

for /f %A in (ssid.txt) do @nsrmm -o recyclable -y -S %A

But, when I run that, it instantly jumps back to a prompt, and doesnt seem to do anything.

Careful - if you delete save sets, nsrmm will delete ALL instances (clones as well, if available).

May I suggest you use the ssid along with the cloneid.