Christopher Faulet

Thanks @idl0r. I agree, it is safer to wait few days before merging the patches. But it remains a good news. At least there is no obvious bug.

@idl0r, Have you experienced new crashes ? If this works, I propose to merge the series.

Ok thanks, I will merge it.

Thanks @idl0r, I'll take a look.

It seems the fix for #2525 (21447b1dd422ec5a2743d9b2affc3a8b5c0fbb48) was shipped with the 3.0-dev8 but not the 2.9.7. So it means 2.9.7 your are not affected by #2525 or it is just...

Another question. Are you using the peer sharding mechanism ?

Sorry for the delay (vacation too).. The 3.0-dev9 contains the fix for #2525 but not the 2.9.7. Only 2.9-HEAD contains it. So if the issue is about 21447b1dd422ec5a2743d9b2affc3a8b5c0fbb48, this should...

Thanks @idl0r. This one is interesting. And it is probably the same issue indeed. It seems `stktable_trash_oldest()` has the same bug that the one fixed by 21447b1dd4. I guess the...

If it is possible on your side, it could be helpful indeed.

So here, we are back again with the first crash. The table's update lock seems to be acquired too long in `peer_send_teachmsgs()` and `process_table_expire()` (and probably in ` stktable_trash_oldest()` too...