Segmentation fault

[hwangumn]

Issue Report

Please describe the issue:

I am running Dorado v0.8.3 on our supercomputing platform with A100 using Slurm Manager. I have successfully run Dorado with smaller POD5 files but once I combined the 24hrs worth of data (PromethION) together. It keeps producing a segmentation fault error with an exit code of 139. Any suggestions?

Logs

[2024-12-11 17:19:14.736] [info] Running: "basecaller" "hac" "/home/ONT/Isolates/raw_pod5" "--kit-name" "SQK-NBD114-96"
[2024-12-11 17:19:17.326] [warning] Unknown certs location for current distribution. If you hit download issues, use the envvar SSL_CERT_FILE to specify the location manually.
[2024-12-11 17:19:17.341] [info] - downloading [email protected] with httplib
[2024-12-11 17:19:17.366] [error] Failed to download [email protected]: SSL server verification failed
[2024-12-11 17:19:17.366] [info] - downloading [email protected] with curl
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed

0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0
17 14.8M 17 2586k 0 0 2728k 0 0:00:05 --:--:-- 0:00:05 2725k
100 14.8M 100 14.8M 0 0 11.4M 0 0:00:01 0:00:01 --:--:-- 11.4M
[2024-12-11 17:19:19.356] [info] Normalised: chunksize 10000 -> 9996
[2024-12-11 17:19:19.356] [info] Normalised: overlap 500 -> 498
[2024-12-11 17:19:19.356] [info] > Creating basecall pipeline
[2024-12-11 17:19:45.588] [info] cuda:0 using chunk size 9996, batch size 5760
[2024-12-11 17:19:46.041] [info] cuda:0 using chunk size 4998, batch size 9216
/var/spool/slurmd/job27986866/slurm_script: line 13: 2146773 Segmentation fault dorado basecaller hac /home/ONT/Isolates/raw_pod5 --kit-name SQK-NBD114-96 > calls.bam

[malton-ont]

Hi @hwangumn,

To be clear, is this segfault occurring with the same data that ran successfully as smaller files?