search

lbcb-sci / raven

De novo genome assembler for long uncorrected reads
MIT License
205 stars 21 forks source link

Raven: stuck at raven::Graph::Construct without any progress. #39

Open shri1984 opened 3 years ago

shri1984 commented 3 years ago

iI, I am trying to run CLR reads (around 50 million reads). It runs for few hours and then simply hangs for hours and hours seemingly not doing anything. At this point, when I run top in command line, job don't show up. when I abort it and resume, it works and then same thing happens again after some time. No out puts other than raven.cereal and a empty out put file. it is happening with subset of original data too. program just hangs.

rvaser commented 3 years ago

Hello, can you paste the log Raven generated? Does it hang on the same part or it changes? How big is the data set and how much RAM does your machine have?

Best regards, Robert

shri1984 commented 3 years ago

Hi, I don't think I got a log file. the information on the terminal for a reduced data set (same output for the full data set): raven::Graph::Construct] minimized 6646090 - 6833623 / 22099221 21.047774s [raven::Graph::Construct] mapped sequences 812.920639s [raven::Graph::Construct] minimized 6833623 - 7021583 / 22099221 20.662825s

In the full data set, I reached the last round of mapping.

When I started the raven with the full data set (50 million reads) it got hung in the beginning after few hours and then I was lucky to resume it from the stage where it hung. Then it hung in the last round of mapping. The same is true for small data set. Of course, I did not run the data set again to check whether it stops at same place. But there is a good chance that it will run for 6 days and stop at the last round of mapping.

The machine has 1.9 TB RAM and I used 220 cores for the run. I just run it with the default settings.

rvaser commented 3 years ago

I suppose 220 cores is an overkill, try with 64. There might be a case where a thread deadlock occurs, but I am not sure what to think.