New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Guarantee failed: [gc_state.current_entry == __null] #2003
Comments
I have a slight suspicion that this has to do with running out of memory and some allocation failing. I previously had crashes with std::bad_alloc() ( |
Just got it again, this time on the existing node instead of the fresh one. I joined a fresh node, marked the previous node dead, waited about 10 seconds and then the existing node crashed with pretty much the same error.
|
Now when I want to restart the node, I get:
|
Crashing a RethinkDB server a couple of times while it's being backfilled to by means of limiting its memory corrupted the table:
I can't restart it even after removing the memory constraint. |
Also I noticed that when a memory allocation fails, we sometimes don't seem to crash immediately (at least in debug mode). For example I got cases like this:
So it ran out of memory, but still had time to run into another assertion. Somehow an allocation failing had side-effects on a different thread, which kept running before the process was eventually terminated. |
... or well I guess std::bad_alloc is sort of expected to cause side-effects because it unwinds the stack. That is bad or course. |
👍 |
... or much easier: Use this function that is part of the C++ standard, to make failed calls to |
A fix for out-of-memory conditions handled incorrectly is in code review 1263 by @AtnNn. However I'm still getting table corruptions. Going to open a new issue for that. |
I got this crash twice when joining a new node to a cluster.
I did something like this:
I might actually have done something slightly different, because I can't reproduce it right now. I also had a compilation job running in the background, so there was some disk and CPU activity.
The text was updated successfully, but these errors were encountered: