The six drives have been moved from one system to another system. In the previous system the partition table looked fine and the raid was unbelievable fast. The new system has an older promise raid controller, but way faster hardware. In the raid promise bios I had to present every disk as a jbod device. Now on the linux boot up every drive shows up. I highly doubt that the promise card made any adjustments to the partition table. If that would be so I would assume that the raid wouldn't even assemble since I am pointing to the entire raw disk. It may be good to reassemble the array with '--freeze-reshape' to allow it to resync without interference from reshaping, hopefully faster, and then '--continue' after resyning has ended. > You think the reshape is so slow right now since a reshape and resync are running? Since the entire system crashed I would assume this is a logical conclusion. Is there a way to double check that I am actually doing both currently?! To start the raid I had to run the export command followed by "mdadm -A /dev/md0 /dev /sd[bcdefg] --verbose --backup-file /root/chunk/1" I assume you want me to do a --stop and then "mdadm -A /dev/md0 /dev /sd[bcdefg] --verbose --backup-file /root/chunk/1 --freeze-reshape" ? At that point I should see the rsync running, correct? Alex On Thu, Dec 20, 2012 at 9:32 AM, Peter Grandi <pg@xxxxxxxxxxxxxxxx> wrote: > >>>> [ ... ] reshape was running it was running around 4mb/s, >>>> which is slow to begin with. [ ... resync too ... ] > >>> A bigger value for 'md/stripe-cache' might help, alleviating >>> some alignment issues, but usually it is not like 10 times >>> better, but might be still worthwhile even if it will be still >>> slow. > >> I tried values up to 32768, but no difference. > > Then probably the partitions are really misaligned. > > Anyhow a wev search for similar cases shows in some cases higher > reshape and resync rates, but not enormously higher, and usually > 'stripe_cache' helps, at least with the reshape. > > Perhaps disabling barriet might help if they are enabled, but > that would be quite dangerous, as you already had a crash during > reshape. > > It may be good to reassemble the array with '--freeze-reshape' > to allow it to resync without interference from reshaping, > hopefully faster, and then '--continue' after resyning has > ended. > -- > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
