Volker discovered "The problem is that we're getting an oplock break but continue to use the write cache.". Patch to follow.
Created attachment 5999 [details] git am patch from Volker for 3.5.next. Volker wrote this, I reviewed and customer reported in the field it fixes the issue. Volker, once you've checked it's the same patch you wrote please re-assign to Karolin for inclusion in 3.5.next. Jeremy.
Yep, it's the right patch. Karolin, please apply for 3.5. The patch also applies the same way to 3.4 and with git am -3 to 3.3 as well, I think a data corruption bug might justify being applied to older releases as well if we ever do one. Volker
Even if we don't do an earlier release than 3.5.x, we can add it to the patches-for-old-versions pages ? Jeremy.
Sure. How does that work? :-) Volker
(In reply to comment #4) > Sure. How does that work? :-) > > Volker > Done. You can have a look at the samba-web commits to retrace.
(In reply to comment #2) > Yep, it's the right patch. Karolin, please apply for 3.5. The patch also > applies the same way to 3.4 and with git am -3 to 3.3 as well, I think a data > corruption bug might justify being applied to older releases as well if we ever > do one. > > Volker > Pushed to v3-5-test and v3-4-test. Closing out bug report. Thanks!