MySQL Bugs: #61208: Possibly invalid assertion in btr_cur_pessimistic

Bug #61208	Possibly invalid assertion in btr_cur_pessimistic_update() inside btr0cur.c
Submitted:	17 May 2011 21:41	Modified:	18 Oct 2012 23:09
Reporter:	Nizameddin Ordulu	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Server: InnoDB Plugin storage engine	Severity:	S3 (Non-critical)
Version:	5.1.52	OS:	Any
Assigned to:	Marko Mäkelä	CPU Architecture:	Any
Tags:	compression, innodb, UPDATE

Description:
This is another obscure bug that happens very rarely. 

Inside file btr0cur.c around line 2294 there is the following assertion:

 ut_a(optim_err != DB_UNDERFLOW);

I think what this assumes is that if btr_cur_optimistic_update() around line 2133 returns DB_UNDERFLOW (meaning that updating the page will make it small) then btr_cur_insert_if_possible() arround line 2270 must absolutely succeed. This is not the case in one of the servers I am testing right now. I have a case where the page compresses really well so btr_cur_optimistic_update() returns DB_UNDERFLOW, but when a record gets updated, the compression rate radically changes so that btr_cur_insert_if_possible() can not insert in place despite reorganizing/recompressing the page, thus the assertion fails. The fix is just to remove the assertion.

How to repeat:
read code.

Suggested fix:
remove assertion ut_a(optim_err != DB_UNDERFLOW)

Another possibility would be to append || page_zip to the assertion. We might also change it to a debug assertion (ut_ad()).

Marko, can you please look at bug 61456 too? I'm trying to make quicklz compression work with mysql and it's very important for me. Thanks for your help.

I associated this bug with the changelog entry for bug#14554000 and several other compression-related bugs with similar effects, rather than creating a new changelog item.