Tahoe Storage Node causes high CPU usage on Marvell Armada 375
Ramakrishnan Muthukrishnan
ram at rkrishnan.org
Mon Nov 14 03:31:07 UTC 2016
On Sun, Nov 13, 2016, at 05:43 PM, exception0x876 . wrote:
> I've installed Tahoe-LAFS storage node on the Marvell Armada 375 box
> (dual-core Cortex A9). As I read in Tahoe-LAFS FAQ, the storage node
> does not require much CPU power due to all CPU heavy stuff being done
> on the gateway node.
>
> However while I transfer the file from the client (gateway is node is
> running on the localhost), tahoe process on the storage node goes up
> to 100%. The transfer speed is around 20Mbps, while I get up to 1Gbps
> using iperf3 between gateway and storage node.
>
> The above makes me believe the storage node still does transit
> encryption/decryption. After the quick look at the sources I assumed
> it uses pycryptopp for this, which in turn uses crypto++. If this
> assumption is correct, crypto++ library is not aware of crypto-dev
> which uses MV_CESA for hardware crypto acceleration. For example,
> openssl is aware of it if compiled with -DHAVE_CRYPTODEV flag and its
> performance is greatly increased than without using crypto-dev.
>
> So the question is, if this is an expected behavior or I am unaware of
> something to do it right?
Could you please try running 'perf top' on a different terminal on the
same machine while you are doing the transfer?
Thanks
Ramakrishnan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20161114/4f3192e0/attachment.html>
More information about the tahoe-dev
mailing list