Monthly Archives: December 2016

System updates

Another year has almost passed and as with most years with leave to clear, this is a period where I update the systems.

The Mac Pro has been developing random shutdowns and is now retired. And its drives removed for other uses.

The server has the most changes, all the internal drives have been upgraded to 6TB drives with the addition of 5 Toshiba 6TB drives. This will provide enough space to house all the data and allow for the retirement of the external eSATA units. The drives on the eSATA units are very very old, some of them are at the 7 year mark. And zfs scrubs are showing lots of checksum errors and there has been one or two data corruptions. The system drive which was a 256GB M.2 has been replaced with the Crucial 1TB SSD from the Mac Pro. The graphics adapter has been changed to Asus NVIDIA GeForce GT 730 which provides for more modern connectors like HDMI and DisplayPort in anticipation of a new monitor. The memory has also been upgraded via 4 x 16GB DDR4 Corsair LPX 3200MHz modules to give a total of 64GB of memory.

The initial idea was to supplement the existing 32GB of memory with another 32GB but the differences in the voltages of the old and new modules meant that this was not possible. Kind of sad.

The disks were replaced one by one, each time physically replacing a single disk and then using zpool replace to bring the pool from degraded to online status before the next replacement is done. This took a very very long time since each replacement took about 5 hours. This is probably a dangerous operation since there is no extra drive during this process and true enough one file had an error which could not be recovered. But since a full backup was done prior to the replacement, the file was restored and then the resilvering done again. This time it well fine.

Once all the drives have been replaced, the next step was to grow the pool to the new size and this was based on instructions at http://blog.ociru.net/2013/09/25/let-your-zfs-extend which worked well. I had a problem with the export which I finally traced to the fact that NFS was still using the pool. Once NFS was stopped, the export worked fine.

The grow operation worked as advertised and the capacity has been doubled. The next step would be to copy all the data on the external eSATA units to the new expanded internal pool. This will probably take a very very long time and I hope that the old disk will hold till the copy is done.

I got a Henge dock for my Macbook Pro Retina 13″ so that it can serve as the Mac Pro replacement and that has been working out very well.

The Windows machine has received a modest upgrade in the form of an additional 16GB of memory bringing its total memory to 24GB.

bonnie++ output for server

The server ZFS pool IO is rather good as shown in the bonnie++ output below:

Version  1.97       ------Sequential Output------ --Sequential Input- --Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
alpha           63G   190  99 529837  74 203121  50   614  97 401102  46 122.7   9
Latency             84706us   24888us     714ms   81362us     851ms     349ms
Version  1.97       ------Sequential Create------ --------Random Create--------
alpha               -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 29949  93 +++++ +++ 21367  71 23405  94 +++++ +++ 27838  90
Latency             14028us     250us   19509us   27425us      30us      97us