V10 Grooming-how long?

[Monafly]

Am I going for a record here? 75 hours since I started a grooming operation and still going. How long do I wait to decide that it's failed and I need to start over?

 

RS 10.01 (105) on Mac mini (10.6.8). Disk media set with about about 4TB of backups on a 6TB drive (didn't observe exact size or file count at start).

 

The last useful message was at 45hrs into the groom "segment 17500 of 18231..." I don't know because of the ellipses whether that was out of 18k or 182k or more segments.

Going home now, we'll see where it's at in the morning.

 

FYI: scripted backups to other sets continue to run during all this time and performance seems typical.

[Maser]

Groom speed is highly dependent on the number of *files* in the media set (and how many past backups you are grooming) -- as well as how fast the CPU is of the computer doing the groom. The *size* of the media set is not as important as how many files are in it.

 

You didn't say what the CPU speed is of your Mac Mini (but I'm guessing if it's running 10.6.8, it's fairly old). CPU speed -- more than RAM or OS version or hard disk speed -- makes the most difference in how fast a Groom will run (all other things being equal...)

 

 

When the groom activity actually displays the "segment X of Y" -- that's when it's actually grooming the media set. If you stop the groom now, you will need to rebuild your catalog file as it will be in an "unfinished groom" state.

 

However, if it never changes the "X" number -- then it sounds like something would be wrong with that and you might have to do that.

[Monafly]

Up to 92 hrs now with no apparent change over the last couple of days. The "segment xxx of xxx" was seen changing at around the 45 hr mark, but since then has never shown it's face. The 'status' portion of the window occasionally toggles between a spinning ball and "Please wait...", but that's all the info I get other than the clock continuing to increase. The log only shows when the groom started-big help.

 

The Mac mini running the engine is a 2Ghz Core 2 Duo with 4GB ram. Not the greatest machine, but should be adequate for RS backups.

[Maser]

if it was me, I'd stop the engine, remove the catalog file for that media set, restart the engine (which will make the groom script fail because it can't find the console) and rebuild the catalog.

 

I can only give a recent comparison of one of my disk media sets:

 

on a 2.5G i5 Mac Mini with 8G RAM, my most recent groom script on a media set that contained about 3M files (600G) -- took 9:15 to run. Your media set (in size) is about 6-7 times bigger. If it has 6-7 times the number of *files* in it, it may take 6-7 times as long to actually run.

[Monafly]

Update: after running for nearly 212 hrs (yes nearly 9 days) I forced RS to quite. Further investigations made me believe that the groom actually was done after a couple of days. This was based on modification date changes in the .rbd files where the actual backup data is stored. Many of the 18231 files (that's how many .rdb files were in the backup set) had change modification dates and were reduced in size from what was typical. Somewhere after all of this 'grooming' was done, the creation of the catalog or something else within RS just wouldn't finish. The last modification date/time on the catalog file was also about 1.5 days after start.

The disk shows significant free space now available, but RS continues to show that very little is available (which was what prompted the groom).

Forcing things to stop meant that the set wasn't usable, so I tried first to repair the catalog (no luck with that) and now started a rebuild of the catalog. That looks like it will take about 24 hrs, so I'll see if things are all back to normal tomorrow.

 

(by the way this set started out with between 4 and 5 million files taking up about 4TB of space)

[Maser]

4-5M files probably shouldn't have taken any more than a couple of days (with your CPU). If you have problems rebuilding the catalog, please let us know (as I'm curious...)

[Monafly]

The catalog did rebuild just fine (took something over 38hrs as it finished during the night). Now seems to be functioning just fine. Space was recovered on the disk. RS properly acknowledges the space on the drive and the script using this media set is working as expected.

 

Next time I have to groom, assuming some update from RS doesn't fix this, I'll look carefully at the date stamps on the actual .rbc files as well as the catalog file and when it appears that they are no longer changing, I'm going to bail on the 'waiting' for the groom to quit, force quit it and just rebuild the catalog. I shouldn't have to do this (are you listening RS?), but at present it appears that's what necessary to accomplish the job.