NASLite Network Attached Storage

Hi,

I shut down my NAS again today, via the shutdown command from telnet, and when it booted back up it started in with the "Checking Storage Filesystem". any idea why that happened? I shut it down like I was supposed to. And now it seems to be taking hours to complete, I have 4x500gb RAID5. If I throw two more disks in there and go RAID 10 would that speed these checks up along with regular file performance?

Thanks,
Jason

Periodically (i.e. every 20 to 30 shutdowns or so) NASLite will check each storage volume (although probably not all at the same time) - this is done as a matter of course and cannot be easily avoided.

You can see the status of your filesystem if you look at the disk-x status pages. When the mount count reaches the maximum mount count, the filesystem will be checked at reboot.

Some experience and tips ...

One of my three disks reached the maximum mount count of 31 (no idea where that number comes from) this morning. Here's a quick summary
Disk-0 300 GB Max mount count 20 check interval 6 months
Disk-1 400 GB Max mount count 35 check interval 6 months
Disk-2 500 GB Max mount count 31 check interval 6 months (actually 15552000 seconds = 180 days)

Of course it was inconvenient, but with the 500GB disk (458GB actual total capacity avail) having 419 GB used by 26,610 files, 2,428 folders, and with the journaling ext3 filesystem, and 1GB ECC memory, 1.2GHz P4-M the file system check only took about 26 minutes to complete. Not bad.

Problem was: my son completed his homework last night before going to bed. And he figured print it before going to school this morning. My systems automatically shutdown at night and restart at 7am. Meaning it was cutting it close ... especially while tapping his foot given the unknown amount of time for the system to be ready this morning ... (now known to be 26 mins for Disk-2)

Since I do shutdown nightly, but occasionally an extra administrative reboot may happen (or a startup might be prevented while out of town), this is what I figured: rather than trying to synchronize the mount count to have file checks only on Sundays (multiple of 7, lets say every 5 or 10 weeks or so) it might be better to use the check interval to make sure it happens on that desired Sunday. For someone who does NOT shutdown regularly (or at least do the one voluntary maintenance reboot on that scheduled Sunday) this plan could become an inconvenience, since any unscheduled boot AFTER that Sunday would require the long start up ...

Just thought I would share some thoughts and the 26 minutes info ... now I am off to investigating how to use LiveCD and the detailed parameters of tune2fs to accomplish the goal.

Georg

P.S.: Additional idea for those who leave their server running 24x7: that one "voluntary Sunday maintenance reboot" could also be automated ... using a WinXP system and Task Scheduler (which is how I start up my NASLite server already anyway) executing a "reboot" login to your server (and/or refer to my SysRC scripts).

I believe that the 31 days corresponds to about one month assuming that you shut down each night.

I leave my machine running 24/7 for literally months on end with no issues. The only time I bring it down is to blow the dust out, say every 6-8 months. This guarantees that the 180 day max between checks is hit and the machine runs a system check upon reboot. Since I dictate the day and time that I bring the system down I am prepared to wait for the machine to become available on the network again.

Personally I think it unwise to shut down a machine like this every since the power and thermal cycling increases the ware on the electronics and mechanical systems of the server.

Just a thought,

Mike

Thanks, Mike, for your additional thoughts ... and I like the "dictatorial" blow-out-dust-and-cobwebs method ... switching 2x per year from/to Daylight Savings time might be an excellent natural time reminder ... although that's not consistently 180 days (the shortest in the near future is 126 days from Nov2008 to Mar2009). (So may be set the interval at 120 or 90 days for 24x7 systems.)

Yes, 31 is a month worth of daily shutdowns ... but why is each disk a different max count # ? I can't be 100%, (more like 99.90), but each of those three disks was formatted by NASLite USB ... why the difference on the three disks (I may have used earlier versions of NL when the respective disk was originally formatted. The new 500GB for sure was version 2.06). I know I have not changed the count with tune2fs before ...

On thermal/power cycling: absolutely correct ... it wears heavier on the electronics. But the counter argument is the waste of power. Especially in the summer, since some cooling is required/wasted ... in the winter the 24x7 "heat" from the servers is beneficial And the 500GB just installed is a new WD GreenPower ... so maybe I should reconsider ... but probably not until I have a suitable and reliable Atom M/B ...

( Mind you I am not a tree-hugger ... but like Mom (or Dad) taught you ... turn off those lights (when not in use) ... and I try to balance that against the increased risk of electronics failure from cycling ... good point you made ... and the benefits of dust-removal )

Georg

I need to do an amp draw test to see just how much power my machine is drawing, I am curious now.

I hate to wait for my machine to come up and so I just let it run, a bit more expensive but....

Mike