NS_ERROR_STORAGE_IOERR: Component returned failure code: 0x80630002

[maxlath]

Using Litewrite, remotestorage was without reason in Offline mode. I clicked the widget to trigger a reconnection and this triggered an error message on the widget (sorry I didn't take a capture) : something like "an error occur" with a button "get me out of here". This last "get me out of here" button triggers a prompt asking if it should erase all the data, an offer I promptly declined.

Then, all remote storage apps seem to suddenly have the same issues with the widget:
on litewrite.net and remotestorage-browser.5apps.com

NS_ERROR_STORAGE_IOERR: SQLite encountered an IO error remotestorage.js:2
NS_ERROR_STORAGE_IOERR: Component returned failure code: 0x80630002 (NS_ERROR_STORAGE_IOERR) [nsIDOMStorage.getItem] backbone.localstorage.js:23

on sharedstuff.org

Exception {message: "Component returned failure code: 0x80630002 (NS_ERROR_STORAGE_IOERR) [nsIDOMStorage.getItem]", result: 2153971714, name: "NS_ERROR_STORAGE_IOERR", filename: "http://sharedstuff.org/js/shared-stuff.min.js?version=8", lineNumber: 52, columnNumber: 0, inner: null, data: null} local-libs.min.js:91

Reloading, emptying the cache or restarting firefox doesn't fix the issue but everything is normal when I use another browser instance.

config: Firefox 29.0 (beta), Ubuntu 12.04 64-bit

any idea what it could be?

Edit: everything back to normal after a reboot. I had unusual disk (i-nodes) issues during the reboot but I don't see how it could relate.

[raucao]

Apparently Firefox's IndexedDB implementation on Linux is not very fault-tolerant. Not sure if we can do anything about that, because we already check for a working IndexedDB on load, and if it's there but throws errors like these, there's not so much we can do. :(

[michielbdejong]

@maxlath the inodes issue will be what manifested itself as the SQLite I/O errors. The same would probably happen if the disk is full, or if hardware errors occur.

We should try to change the 'an error occurred' message to something mentioning the device hard disk in this case.

Thanks for reporting!

[raucao]

We should try to change the 'an error occurred' message to something mentioning the device hard disk in this case.

I don't think that's possible when Firefox craps out with native exceptions. Would be cool if we could catch it, but if it doesn't throw an appropriate JS exception, we're out of luck.