Hi Alexander, On 11/09/2016 07:45 PM, Alexander Barkov wrote:

Hello Elena,

On 11/09/2016 08:38 PM, Elena Stepanova wrote:

...

Hi Bar, Sanja,

On 11/09/2016 06:28 PM, Alexander Barkov wrote:

...

Hello Sanja, Elena, Wlad,

Sanja, please review a patch for MDEV-10780.

The patch is for 10.1. I could not reproduce it in 10.0. But it should be safe to push it into 10.0 anyway.

Yes, please push into 10.0 if it's safe enough, I am still getting it rather easily on the current 10.0 tree by running the test from the description with --repeat=N (I run with N==100, but it actually crashes on the 2nd iteration at the moment).

Okey.

Can you please check if it fixes MDEV-10910 and MDEV-10806 as well?

I've checked that the patch fixes MDEV-10806 and closed it as a duplicate. There is no decent test case for MDEV-10910 (there was no point investigating it because we already had two similar issues), that's why it's still "draft" and assigned to me. It's just something that has been happening in my tests sporadically. I'll keep observing for a while, and if it doesn't re-occur after the patch, I'll close it. Thanks, Elena

...

Regards, /E

...

This patch is most likely fixing MDEV-10806 (assigned to Sanja) and MDEV-10910 (assigned to Elena).

Thanks!

Hello Sanja, On 11/10/2016 01:13 AM, Oleksandr Byelkin wrote:

Hi, Alexander!

On 09.11.2016 16:28, Alexander Barkov wrote:

...

Hello Sanja, Elena, Wlad,

Sanja, please review a patch for MDEV-10780.

The patch is for 10.1. I could not reproduce it in 10.0. But it should be safe to push it into 10.0 anyway.

This patch is most likely fixing MDEV-10806 (assigned to Sanja) and MDEV-10910 (assigned to Elena).

The patch is good, but I definitely want to see commit comment (it is also important part of the fix). Could you send it, of commit and forward me the commit e-mail?

Here's an updated version with a comment and with a very small addition fixing another bugs. Thanks!

[skip]