On Fri, Mar 13, 2015 at 02:48:03PM +0000, Adrian Bridgett wrote: > Slow as in sync(1) taking 6 seconds on an idle system, dd-ing from the raw > disk drops from 600MB/s to less than 1MB/s. iotop shows jbd2 often waiting > a long time, but _very_ low throughput (KB/s it that). I've spent a day or > so trying everything I can find to identify the cause but I'm completely > stuck (even with help from a btrfs guru - he eventually suggested asking > here). The obvious thing to do is to try using blktrace to see which I/O operations are triggering the problem. It may just be that the hard drive's internal FTL metadata is very badly fragmented. If so, it could be that that the only thing that will fix things is to reset the entire FTL is to do a full backup, do a discard or a secure erase on the entire SSD, and then reformat and do a restore from backups. You might also try looking at the SSD model and try to find some reviews on Tom's Hardware or Anand Tech to see how the SSD fares when stressed on various SSD torture tests. What model SSD is it, by the way? - Ted -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
