Song re-starts at about 8%

[zlurg]

I wish like crazy that I had a log to show, or a screenshot, or if I could make this happen on command, but I'll log it here for posterity just in case an investigation method is found.

SOMETIMES...when playing on shuffle usually, a FLAC file will just randomly restart at less than ten percent.  The progress scroll doesn't go backward, the song just re-begins.  No gap, no delay, no SFX, it just restarts.  Does not record a log entry that I can find.

Any ideas how best to document this so I can ask properly for help?

[Steven]

it could be multiple things
- check the settings tab in the tag editor - do you have remember position ticked
- is skip starting/ending silence ticked in Preferences/ Player
- do you have dsp plugins enabled

if its none of those then send a link to one of the files

[zlurg]

It's none of those, so we at least know it's unusual.  No plugins, no preferences, and it's very random, like maybe 10 times a week, assuming I use the player 6-8 hours a day.  One day, it happened 3 times in one setting.

The first couple times this happened, I thought the FLAC file was corrupt, so I hit "back" and played it again, but it NEVER skips twice in a row, which makes this tough to debug.

You would know better than I, obviously, but it "seems" like it's a cache or buffer issue, and something is tricking the player.  What I believe is that I'm doing something on the PC in my absentmindedness and it's causing the issue.  The thing is, this PC has 64GB of RAM, and I don't do anything intense on it, no games or CAD or video, but I concede it's probably me making a condition that manifests as a random FLAC starting over after a few seconds.

[phred]

if its none of those then send a link to one of the files

[zlurg]

What do you want, a FLAC file that won't repeat behaviour on demand?  I can't get the problem to repeat, so I seriously doubt it's the file itself.

Here's another clue: as near as I can tell, it only happens when the playlist is on shuffle.
More details: 250k tracks in the library
                        Playlists of this type are usually >100 tracks
                        I sometimes get a different error about permissions,  but that's another topic.  That one I usually fix with attrib in the directory itself.

[zlurg]

Sorry, one more thing: I did install the 3.5 patch earlier this month, but the problem has been going on at least since August.

[zlurg]

You know, I'm starting to feel kinda stupid.  If I could just collect my thoughts and put them in a single post, I'd be a nicer neighbor.  But I wanted to add this:

MusicBee itself is installed on the boot drive (Samsung Evo 2TB M.2 SSD).
The music library and database, however, are on an 8TB SATA drive
Motherboard is an Asus Z10PE-D8WS
Xeon E5-1620v4 @ 3.5GGHz
1080ti graphics card
4 SATA rotating rust on the SATA bus, M.2 on the mobo
Windows 10 Pro, 22H2 19045:3155
No RAID

[Steven]

If its not specific files then i dont think i can help as I cant recall anyone ever reporting this issue before.
The only suggestion i have is perhaps the drive is sleeping and somehow the spin wakeup affects loading the start of the music file

[zlurg]

I understand.  I wish it would log something, but there's not even a notice that it's about to happen, it just does it.

My guess is similar, and probably is the difference between the M.2 and a different SATA drive, like maybe something in the limited-rewrite fetaure of an SSD, or maybe the program thinks a song 30 plays away should start at xx:xx as a checksum, but the actual speed of the drive exceeds that estimate so it's off by a few seconds?  I dunno.  It's strange, to be sure, and if I could just get it to consistently report, we'll all be better off. 

I guess I'll keep scrambling for clues, but I do think it might be time to move MusicBee over to that same HDD as the library and see if the problem still exists.

I'm also intrigued by your suggestion, like if it's an issue with the random() functioning in the shuffle.  Maybe what I'm seeing is the gap between the program selecting the next song and the HDD delivering the rest of the song from a quiesced HDD.  If the program needs a few seconds of audio file to provide gapless/crossfaded playback, that's a different file read than playing the full music track, yes?

OK, thanks for the ideas.  I'll be sure my next few reports have a more thorough account in the request.

[hiccup]

I would suggest to keep MusicBee on the ssd, and (as a test) only play music that is also on that ssd. (copy some of your music to it)
Also see if you can adjust the go-to-sleep time to a longer period for the hdd that currently contains your music. (or even disable sleep mode for it, as a test)

[zlurg]

Yeah, some good ideas here, so I did a little investigating, and I think you're on the right track, @hiccup

It seems to me that MB is building the playlist on the SSD's temp directory as soon as I hit play, and may even know what order it'll play tracks, but in the time it takes to get to the song that skips, the source media HDD has gone to sleep and it takes a few seconds for the drive to spin up.  Just a guess, but it's as good as any.

I checked the power settings for that drive, and its factory setting is 20 minutes, which was surprising.  I changed it to 60 minutes and set a timer.  It has passed 20 minutes and hasn't repeated behavior.  When it gets to 60, we'll see if this is predictable.  If not, it was still nice to learn my HDD power settings were utter kr@pp.

[hiccup]

Yeah, some good ideas here, so I did a little investigating, and I think you're on the right track, @hiccup

Credits go to Steven for first suggesting that.

[zlurg]

Still happens after 60 minutes of continuous play, which means this is possibly reproducible now.

I will now try setting sleep time to "Never" on that library drive.

Smoking gun?