#1170 closed defect

does new-downloader perform badly for certain situations (such as today's Test Grid)? — at Initial Version

Reported by: zooko Owned by:
Priority: critical Milestone: 1.8.0
Component: code-network Version: 1.8β
Keywords: immutable download performance regression Cc:
Launchpad Bug:

Description

I tried to download a movie file from Test Grid with Tahoe-LAFS v1.8.0c2 (new-downloader) and the file wouldn't download fast enough or smoothly enough to watch in a streaming fashion during download.

If you look at the download status that I posted on #1169: http://tahoe-lafs.org/trac/tahoe-lafs/attachment/ticket/1169/down-0.html , you'll see that it often took a second or two to fetch a single segment. Also, I don't exactly understand the meaning of the fields (hence #1169), but it looks like downloader was scheduling up a large number of 32-byte reads. Is that right? Were those getting coalesced before sending them?

It could just be that the current test grid has servers that take seconds to respond to requests and downloader can't do much about that. Is that what happened? Could downloader have done something about it, such as by choosing different servers instead of the ones that responded so slowly?

I guess I should compare old-downloader to new-downloader behavior on the Test Grid, but right now I have to go to sleep!

Oh, other people could do that comparison with this file if they like:

http://localhost:3456/file/URI%3ACHK%3A4klgnafrwsm2nx3bqy24ygac5a%3Acrz7nhthi4bevzrug6xwgif2vhiacp7wk2cfmjutoz2ns3w45qza%3A3%3A10%3A1490710513/@@named=/bbb-360p24.i420.lossless.drc.ogg.fixed.ogg%2Bbbb-24fps.flac.via-ffmpeg.ogg

Change History (0)

Note: See TracTickets for help on using tickets.