Open jonblack opened 7 years ago
We've been having this segfault for a long time now, without knowing where it comes from. It's quite annoying. So I think the answer is yes, we should improve logging.
We've added more log statements now, telling which request mrs was handling at the moment it segfaulted. However, there is not a specific web request that co-occurs with the segfaults, so it doesn't say much. Also, MRS is multi threaded.
What would help more is having core dumps with stack traces in it. See also: #66
The following error appears in the logs:
There is no additional information, so it's very difficult to figure out the cause. See how far we can get, if it proves difficult, we can open up an issue to improve the logging.