[OpenAFS] mounting lots of volumes quickly

[Hartmut Reuter]

Leland J. Steinke schrieb:
> Neulinger, Nathan wrote:
> 
>> Did you build with the Bitmap-later option? That will also improve
>> startup speed quite a bit.
>>
>> Our servers can usually restart in 2-3 minutes. Though they generally
>> only have a thousand or so volumes on each one.
> 
> 
> Actually, I set both --enable-fast-restart and --enable-bitmap-later 
> when I ran configure, but this was after I had already built the 
> volumes.  That is the problem with reading the QuickStartUnix and 
> actually typing in what one reads! If I need to rebuild the volumes with 
> these options enabled, so be it.

There is nothing special with the volumes if you run your server built 
with --enable-fast-restart and --enable-bitmap-later. Fast restart only 
skips salvaging after a crash which can save much time, but only after a 
fileserver or machine crash.Bitmap later means that the bitmap of used 
and free vnodes in main memory is built not at start time of the 
fileserver but only when its needed the first time. You can switch at 
any time between fileserver binaries compiled with or without these options.

Hartmut


> 
> On the down side, your experience of (2-3 minutes/1000 volumes) times 
> the ~30000 volumes I am trying to mount comes up with a start time of 
> 60-90 minutes, which is more than what I am seeing (40 minutes) :-(
> 
> Thanks,
> Leland
> 
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info


-- 
-----------------------------------------------------------------
Hartmut Reuter                           e-mail reuter@rzg.mpg.de
					   phone +49-89-3299-1328
RZG (Rechenzentrum Garching)               fax   +49-89-3299-1301
Computing Center of the Max-Planck-Gesellschaft (MPG) and the
Institut fuer Plasmaphysik (IPP)
-----------------------------------------------------------------