Good afternoon! My question: Can I leave writeback_running=0, or will it hurt my data? Background: I've started using bcache on my NAS array and must say it is working out quite well. My goals were energy/noise savings and performance. These appear to be met with my current configuration. backing device: RAID5 array on 4x WD green drives cache device: RAID1 array on 2x SanDisk SSDs LVM on top of bcache0 WD drives set to spin down after 10 minutes idle linux 3.12 writeback_percent = 25 sequential_cutoff = 0 congested_*_threshold_us = 0 writeback_running = 0 ... with these settings my server can happily continue running its periodic tasks (smartd, cacti, syslog, dvb, samba, ntp, etc) without spinning up the disks. All periodic data gets stored to the energy saving (and silent) SSDs. Great! My concern is what will happen if some huge burst of IO comes and overwhelms my cache drive. Can bcache still push data to the RAID5 when writeback_running=0? I have a daily cronjob set to re-enable writeback_running and set cache_mode to writethrough, wait for clean, then re-enable writeback. So, the dirty_data does not tend to get too high. However, there could be some day where a ton of data gets written. Is near-permanent writeback_running=0 safe? Thanks! PS. Obviously I am interested in the progress of RAID5-aware bcache and multiple SSDs per cache set, since both would directly benefit my setup. So I am pretty happy with the direction bcache is heading! -- To unsubscribe from this list: send the line "unsubscribe linux-bcache" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
