Dennis Nienhüser
Dennis Nienhüser
Thoughts, anyone?
Same here with mlflow 2.4.0 (client) and mlflow 2.4.1 (server) and a file of 830 MB. Any insights or workarounds?
> Same here with mlflow 2.4.0 (client) and mlflow 2.4.1 (server) and a file of 830 MB. Any insights or workarounds? You can work around it by replacing the call...
Did you try the workaround from https://github.com/mlflow/mlflow/issues/7564#issuecomment-1717949904 yet?
> The **iter_timestamp is in seconds (of type double) but ros::Time::fromNSec(uint64_t) expects nanoseconds (of type uint64_t) which caused segfault in analyzePoints. Bug fixed by using ros::Time(double) to get point_time. Thanks...
Can you provide more details? E.g. more log output, what have you been doing before the crash occurred, does it happen every time or just once?
Great, this helps a lot. This error usually occurs when creating ros::Time instances with invalid values. I have fixed similar errors before - seems I didn't catch all yet.
Thanks for tracking it down. I wonder if context.time or point_time is_zero() is true here. This would cause the error and would require a special case also (probably do not...