wiki:TestGrid

Version 24 (modified by warner, at 2008-03-28T22:51:37Z) (diff)

add new graphs, remove old dirnodes one

The Public Test Grid

How To Connect To The Public Test Grid

The test grid is subject to being updated at random times, so compatibility is likely to break without notice. Also note that many upgrades require all files and directories to be flushed, especially as storage formats change.

The test grid is currently running an 1.0.0-compatible release (currently 1.0.0-r2366).

After installing the code according to docs/install.html, copy the following file into its base directory:

file namefile contents
introducer.furlpb://todjw7qkb4dgq4fkeo7cqydcu5vneioh@tahoecs2.allmydata.com:52106/introducer

Then run tahoe start.

We created a shared public directory: feel free to use it for experimentation -- once your node is up and running on port 8123, this URL should give you access to that directory.

If you aren't running a node locally, you can replace "127.0.0.1" with another host, such as "tahoebs1.allmydata.com" in that URL, like this: URL to the publically writeable directory through tahoebs1.allmydata.com.

Test Grid Statistics

Usage statistics are available for the public test grid. NOTE: these are probably out of date and inaccurate. #13 and other tickets are where we're keeping track of better grid-wide instrumentation tools that we want to build.

These measurements have been broken by changes to the Tahoe source code.

Count of how many clients and servers are connected to the grid

http://allmydata.org/tahoe-munin/tahoecs2.allmydata.com-tahoe_introstats-month.png

Load Average

This measures how long a 1 second callback is overdue, on each machine that connects to the central 'stats gatherer'.

http://allmydata.org/tahoe-munin/tahoecs2.allmydata.com-tahoe_runtime_load_avg-month.png

Same, but those the peak value (reset at node reboot)

http://allmydata.org/tahoe-munin/tahoecs2.allmydata.com-tahoe_runtime_load_peak-month.png

Total space consumed in grid

This queries each storage server for how much space it is currently consuming.

http://allmydata.org/tahoe-munin/tahoecs2.allmydata.com-tahoe_storage_consumed-month.png

The following counts are samples of a single storage server, running four nodes. The test grid uses four servers, running a total of 13 nodes.

count of files stored in the grid

http://allmydata.org/tahoe-munin/tahoebs3.allmydata.com-tahoe_files-month.png

average shares per file

this hints at the number of nodes in the overall grid; If the number of hosts in the grid is less than 10, then shares_per_file should equal about 10/numhosts

http://allmydata.org/tahoe-munin/tahoebs3.allmydata.com-tahoe_sharesperfile-month.png

total bytes consumed by all of the Tahoe test grid storage servers operated by Allmydata Inc.

http://allmydata.org/tahoe-munin/tahoebs3.allmydata.com-tahoe_storagespace-month.png