2009-12-20 05:17 AM
I have a 2TB volume that I use with NFS for VMWare. This volume had 720GB used before I ran dedup and now it has 320GB, a nice space savings. I kicked off dedup manually at 4pm and kept an eye on it wanting to know when it completed. By issuing the sis status command I found it took 4.5 hours to run. I assumed that it would take less time the second time around so I kicked it off manually again and this time it took about 2 hours.
I now have it set to kickoff each evening at 9pm. I want to make sure it has completed before I run SMVI to take snapshots and run my Snapmirror. I can assume it will take 2 hours to run dedup and schedule my SMVI jobs accordingly. Is there any way to tell how long the dedup job that ran the night before took? Is there any log file I can look at? Is there any way to set an alert that sends me an email giving me this detail?
I am running SMVI 2.0 and OnTap 7.3.2 on my FAS3020 filers
2009-12-21 03:03 AM
Thanks, I will take a look in the log. I did find out one thing that is interesting. Last night at 9pm I was logged into my filers just to make sure that the dedup jobbed kicked off fine. I issued the sis status command a few times and noticed that the dedup job that I kicked off manually that took about 2 hours only took about 10 minutes when scheduled. In other words, issuing the command:
sis start -s <vol>
is different than having dedup start when it is scheduled. If I remember correctly, the sis start command scans all the data first and then goes back and does a search for blocks that can be deduped. It appears that if this process is started with a schedule, only the search step needs to be done and it is very fast.
2009-12-21 06:33 AM
Yes it is!! The -s switch forces a complete rescan of the data in the volume while the normally scheduled job will look at just the new blocks since the last pass. That is the normal behavior. the -s is really only needed on the first pass. You should only need 10min of so each night...