[BUG] Jaeger take more cpu than asked

[rdenise]

Dear authors,

I am currently deploying Jaeger on a cluster to distribute analysis tasks. However, I have encountered an issue during execution. After the scanning for terminal repeats step, the process outputs Using OneDeviceStrategy [‘CPU:0’] and subsequently appears to attempt to utilize all available CPU cores on the machine. This behavior results in a bug that prevents proper resource allocation.

For your reference, I am running Jaeger with the following command:

jaeger run -i {input.contigs} -o {params.jeager_folder} --cpu --getalllabels --overwrite --workers {threads} --prophage --getsequences -f

Could you please advise if there is a parameter, environment variable, or alternative configuration that can restrict Jaeger’s CPU usage to a defined number of cores? Any guidance or suggestions to resolve this issue would be greatly appreciated.

Thank you in advance for your help and for your continued efforts in developing Jaeger.

[Yasas1994]

Dear @rdenise ,

Thanks for reporting this bug. We will fix this in the next release.

For now, please set up these environmental variables before running Jaeger and see if that helps.

TF_NUM_INTEROP_THREADS = {num_threads}
TF_NUM_INTRAOP_THREADS = {num_threads}
OMP_NUM_THREADS = {num_threads}