Domino on Linux/Unix, Troubleshooting, Best Practices, Tips and more ...


Daniel Nashed

Avoiding disk fragmentation via nshrun

Daniel Nashed – 25 May 2010 04:52:44

Some of my customers are already running optimized compact operations to avoid fragmentation of their data disks with NSF.
They use a combination of compact -B and -b with the -S option to avoid recovering small amount of space from databases.

My tool nshrun already has an option to decide which compact to use based on the free space in a database.
If just a bit of space is free no compact is invoked. If just some space is free a compact -b is used if much space is free a compact -B is used.
Already in previous versions you could specify those options using -S and -s in nshrun.

But this does only reduce fragmentation to some extend. I would wish there would be an option for compact to keep some free space in the database.
There is currently no way to keep some space in the database. But I found a way to allocate some bulk space in the database in one piece which could avoid smaller allocations in the NSF would create many smaller disk fragments to be allocated. So you now can specify an amount of free space that should be in the database after compact finishes and frees space in the database.
This amount of free space is allocated in one step and would only result in at most one more fragment.

The new option is currently under test but if you are a nshrun user I am happy to send you a new version with this option to check if this helps to reduce your NSF disk fragmentation.
It will also not completely resolve all fragmentation but could potentially reduce the fragmentation a lot.

I would be also very interested to get your feedback about disk fragmentation thru NSF. I personally think there is huge potential in avoiding fragmentation.
This is not primary for Domino run-time performance but more for backup performance.

-- Daniel



    • [HCL Domino]
    • [Domino on Linux]
    • [Nash!Com]
    • [Daniel Nashed]