subversion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Philip Martin <phi...@codematters.co.uk>
Subject Re: Potential regression: high server-side memory consumption during import
Date Sat, 03 Mar 2018 13:06:49 GMT
Stefan Sperling <stsp@elego.de> writes:

> I may have found a way to store an svn_repos_t object on the connection.
> Does it help? (authz_tests pass)

No significant improvement.

I am timing

  time svn log http://localhost:8888/repos/asf/subversion \
     --username pm --password mp > /dev/null

and I realised my earlier timings had the apache LogLevel set to Debug
which means apache spends a significant time logging.  Lowering the
LogLevel to Crit improves the speed.  Timings for 7 runs after starting
apache:

         trunk      patched     reverted
          1.11         1.11         1.11        1.9

          5.2s         5.1s         3.1s        7.9s
          3.4s         3.4s         1.3s        4.5s
          3.7s         3.8s         1.7s        4.5s
          3.9s         3.9s         1.8s        4.5s
          4.0s         4.0s         1.9s        4.5s
          3.9s         3.9s         1.8s        4.5s
          3.8s         3.8s         1.8s        4.5s

Now that apache is logging less the performance hit is even clearer.
1.11 could be 2 to 3 times faster than 1.9 but the authz effect means it
is only about 10% faster.

Note an odd effect in the above numbers.  The second run for 1.11 is
always the fastest. The first run in each case is the slowest, the
Subversion cache is cold.  The second run is faster, the Subversion
cache is hot.  In 1.9 subsequent runs are comparable to the second, but
in 1.11 the subsequent runs are a bit slower than the second.  I don't
know what causes this effect.

-- 
Philip

Mime
View raw message