Hello,

Does anyone here run a /usr/portage mounted over NFS
over an ADSL link, like 1.5x256?

I do, and it's incredibly slow. I have an
`emerge --update --deep --pretend world` going that
has been running for over a half hour now. I have
about 600 packages installed. And the network
throughput, since this NFS link is between two ADSL
1.5x256 connections, is limited by the 256k upload
speed. Again, this is just a --pretend. I'm not even
downloading packages at this point.

I think this is a good argument for some sort of centralized
portage hash database. I think portage should try hard to
avoid having to search the filesystem in a O(N) manner
every time an emerge operation is run. This info should
be cached in a database somewhere so that retrievals
can be swift, even over low bandwidth NFS links. The hash
database should be incrementally updated after every non
--pretend emerge operation and completely rebuilt from a
complete portage filesystem scan after each `emerge sync`,
and possibly completely rebuilt each night from a cron job.

I think DJB's CDB database format would be a good choice
for such a hash database, but then again I don't know
what kind of data we would be caching. What is portage
doing during that 35 minute `emerge --update --deep
--pretend world`? Please tell me that it isn't reading
each individual ebuild file...

-- 
Jesse Guardiani, Systems Administrator
WingNET Internet Services,
P.O. Box 2605 // Cleveland, TN 37320-2605
423-559-LINK (v)  423-559-5145 (f)
http://www.wingnet.net



--
[email protected] mailing list

Reply via email to