Post subject: Stability "burn in" test of svn-1171

[Ted Harper]

(Also posted in rokulabs Firefly forum)

I posted the following a couple of days ago (in a thread on the rokulabs site), and thought I’d follow up now using svn-1171:

Quote:
What I found (with the Windows version of the previous nightly before 1158 – haven’t run 1158 enough to know if it’s different) is that after some period of playing music via RSP to my M1001, the Soundbridge loses the ability to talk to anything on the network until it is physically powered-off-and-on (ie even a reboot via the menu doesn’t make it able to work properly).

This is with 2.5 beta 10 firmware and doesn’t seem to happen when using UPnP AV as the streaming protocol (from Twonkyvision) or even DAAP from Firefly, just would appear to me to be something in RSP – on either the Firefly side or the Soundbridge – which wasn’t happening with the early Firefly RSP version. If it matters almost all of my music is WMA (CBR or VBR, no lossless), and this is what was being played every time I have had the effect I described here.

Using svn-1171 (same server PC, same music library, same firmware on the Soundbridge, etc), I have been _unable_ to reproduce the Soundbridge failures as described above and which I could reproduce sooner-or-later with svn-1130.

I played several albums through start-to-finish with no problems, then last night I left the whole library on shuffle-play for 12 hours non-stop and it was still playing this morning. As well as wanting to see if it recur on the new build, I wanted to do a “burn in” test of the whole environment (ie wireless connection, memory leaks anywhere, etc) to make sure it was indeed stable for long-term continuous playback, and it seems to be fine.

I guess it’s unfortunate I didn’t get a crash or lockup on either the Soundbridge or Firefly service, as far as tracking down the problem other people are reporting, but I thought I’d better post that I _didn’t_ get a crash in my environment. Specifically my music library is about 5500 tracks, about 90% WMA (128-200kbps – ripped from CDs over the last few years and the most recent ones in VBR) and 10% MP3; no DRM on any of them.

If the Soundbridge failure problem I described earlier with svn-1130 and quoted above _does_ recur for me now (ie after it didn’t happen in 12+ hours of testing of svn-1171), I don’t know what to do to narrow it down or provide Ron with an easy repro scenario for it. Fingers crossed it just doesn’t happen anymore with svn-1171 – thanks Ron if you fixed it, but I didn’t see anything in the bullet points on the nightly update that explicitly described a fix in that area.

ted.h.

[rpedde]

@Ted Harper wrote:

I guess it’s unfortunate I didn’t get a crash or lockup on either the Soundbridge or Firefly service, as far as tracking down the problem other people are reporting, but I thought I’d better post that I _didn’t_ get a crash in my environment. Specifically my music library is about 5500 tracks, about 90% WMA (128-200kbps – ripped from CDs over the last few years and the most recent ones in VBR) and 10% MP3; no DRM on any of them.

It may be specific to something I can’t figure — only transcoding certain types of files, or files encoded by a particular encoder, heck, I dunno.

I don’t rigorously test this stuff — these are nightlies, and I figure some instability is par for the course. I do run a short stress on them, do browsing and playback of a hundred or so songs before I release them, so it’s usually a reasonable test that it’s not completely broken, but there might be any number of cases with different versions of ffmpeg, or zlib, or any number of other things.

I’m still hoping I can repro some of the stuff people have reported, because it will make it much easier to nail down if I can run a debugger against it. Otherwise, it’s going to be trial and error, and will take some time.

Anyway, glad to hear this report. Thanks.

— Ron

[Author]

Posts