Interrupted analyses on long runs with Beast v.1.10.4

[Alex Byrne]

Hi all,

I've been running some long (3 weeks plus) analyses using Beast v.1.10.4 on a cluster in my institute. I knew the runs would take a long time and that it could be interrupted so I made sure to include the '-save_every' option to generate regular state files (~every 24 hours). My analyses were then interrupted due to maintenance on the cluster and my runs were ended before they had completed. I've now tried to restart these analyses using the '-load_state' option with the latest state file, but it appears that the iterations begin again at 0, rather than picking up from where the interrupted runs got to (a few million iterations). Is this normal behaviour or should the restarted analyses pick up from where the interrupted analyses got to?

Any advice would be greatly appreciated.

Thanks,

Alex