I've had a problem with galaxy all day, when I try to use the BWA-MEM tool I get the message "This job is waiting to run" but it never runs. I've tried it as part of a workflow and as a stand alone test and get the same. The fastQC and remove sequence artifacts tools appear to be running fine. I've even tried using data I've previously used to do a BWA-MEM on Galaxy just in case! Does anyone know what might be happening?
The public Main server is very busy and jobs may be delayed at this time. Jobs process in the order submitted. This is true both for your own jobs (same history or across histories) and jobs submitted by other users. Jobs are balanced so that each user has equal queue access per job and by job type.
In simple terms this means that if one user starts 20 jobs and another 2, the first job submitted by both will enter the queue. Then once that job completes, the next round of job queuing begins for each subsequent job started per user, in the order of submission. If a user starts multiple jobs, the wait for all to complete (as a whole) will be longer (the more jobs, the longer the total processing time). This is the same whether there were many jobs initiated with a workflow or many jobs serially initiated from within a history. It is definitely a good strategy to queue jobs that are related to the same analysis (using a workflow or otherwise) as soon as possible, to get them into the queue.
If a computationally intense job is submitted by a user, the wait for the next job of that type will have a longer wait time (the first must complete before the next is executed). But jobs of a different type may queue and execute if there are no upstream dependencies. This is because jobs of different types are sent to different clusters depending on the resources required. Computationally intensive jobs tend to have longer wait times when the cluster that processes them is busy, due to each larger job running longer than simple jobs sent to the smaller cluster (which tends to have a shorter queue time because each individual job executes quicker).
In almost every case, it is best to allow queued jobs to stay queued to retain the original position. Stopping and restarting jobs moves them back to the end of the queue. That said, if some work is more important to complete quickly, waiting to initiate other work until that first batch is completed can help. If you do delete some jobs in order to prioritize others, do not restart them until the other priority work is completed, or your own jobs could compete against each other for queue space (effectively making each workflow/set of jobs take longer to complete as a whole).
More about job processing: Support#Dataset_status_and_how_jobs_execute
Our apologies for the current wait time. Significant resources are dedicated to job processing, but at times demand can exceed that available resource.
Thank you for your patience, Jen, Galaxy team