Closed drzhouq closed 3 years ago
This is an issue with s2p, who's calling mgm. Abort is sent to mgm probably because of a timeout.
@gfacciol our mgm_timeout was set to 24 hours, which obscenely long. We were running s2p process on a 96 vcpus / 512G AWS EC2. But the issue is moot since the preferred algo is mgm_multi, we will try to get that work. It seems to us that mgm produces better results than mgm_multi, for some unknown reason
Thanks for the feedback! I'd greatly appreciate if you have an example, that you can share, showing the difference between mgm and mgm_multi. It would help in finding explaination.
when running mgm, we got random sigabbt error. core dumps are also been generated when error take place