Bugzilla – Bug 2595
Spurious disk-full errors may happen during recording
Last modified: 2020-11-30 05:51:06 UTC
My debugger stopped just at the point of throwing the exception in the checkpoint thread. A closer look showed that the error from sqlite was SQLITE_BUSY and that the main thread was also in the middle of sqlite function calls, only to perform a read transaction while redrawing the screen.
Created attachment 1020 [details] waveform lagging Testing on W10 with Audacity 3.0.0 045bedf (non-debug build) Using a USB clip drive FAT formatted with 12GB free space The recording is painful to watch - the waveform display lags a long way behind the Timer toolbar (see attachment) Several minutes further on and after a couple of zoom outs to try to see more this is still sort of progressing (no spurious disk-full).
Created attachment 1021 [details] after hon hour After an hour with no spurious disk-full message this was the state (see attachment). So then I pressed the Stop button - still the Timer and the Recording meter carried on for a while then a grayed=out Audacity window, which yielded a Blue-Circle-Of-Death when I clicked in it,
Created attachment 1022 [details] Grayed out after clicking Stop Grayed out Audacity after Clicking the Stop button - Blue-Circle-Of-Death present too but not captured by "prt sc" keyboard button. Note the Time reset to zero indicating that the Sop has happened.
Created attachment 1023 [details] failed Recovery After 20 minutes of "Audacity not responding" in the Task Manager I crashed the process tree. When I restarted Audacity I was offered Recovery of the project - but recovery failed with >File Error >Audacity failed to read from a file in G: Plus I get left with a garbage 1.6 GB file on my USB thumb drive
Testing on W10 with Audacity 3.0.0 045bedf (non-debug build) Recording to my onboard D: (spinning metal) drive with 0.66TB free on my 256 SSD HP laptop. Project rate set to the max 384000 - recorded fine for an hour, no spurious disk-full error message(s)
(In reply to Peter Sampson from comment #5) So you failed to reproduce this moonphase in the commit previous to my fix. Are you willing to accept that I know what I saw on a Mac build and made a real fix for it?
The fix for the problem I know I saw is at e00f8da5d5dd45542d101f2e4dc7476c4089c5e3
(In reply to Paul L from comment #6) >Are you willing to accept that I know what I saw on a Mac build >and made a real fix for it? Yepper - black-box testing moonphasers like this is like "nailing jello to a wall" ;-)
Closed on the basis of Paul's Comment #6
