Parity check on every reboot

[Flibblebot]

Hope someone can help me.

 

I'm running unRAID 4.7 and have a cron job setup (via the go script) to powerdown (using the powerdown script) at 23:07. The powerstrip supplying the server is powered off using a timer switch at 23:30 and powered back on next morning at 08:00, with the BIOS set up to boot on power restoration.

 

This works fine, but the main problem is that unRAID goes through a parity check on every reboot - even though the array is clean and always completes without any errors.

 

The unRAID install is vanilla apart from the S3 sleep script (I get another server to wake the unRAID server up at 22:55)

 

I'm at a loss to see where I should be looking to find out why unRAID insists on performing a parity check every boot.

syslog-231111.txt

[graywolf]

Since you said it is Vanilla Setup, sounds like you have the original powerdown script that does NOT kill processes holding the mounts open and therefore unRaid is not shutting down cleanly.

 

That is the reason for the parity check every time.

 

There is another powerdown script (i installed it using unmenu) which will kill connections that are accessing the disks so that unraid will powerdown cleanly.

 

Clean Powerdown Currently Installed.

Will be automatically Re-Installed upon Re-Boot.

Description: Primary purpose is to provide graceful shutdown of the environment when called from scripts or via command line. Sets ctrl-alt-del to do powerdown instead of reboot

 

The lime-technology stock version of powerdown will be renamed as unraid_powerdown if it exists.

Package URL: http://unraid-powercontrol.googlecode.com/files/powerdown-1.02-noarch-unRAID.tgz

Package File: powerdown-1.02-noarch-unRAID.tgz

md5 Checksum: 10eff6bbb4a2f1428b4c57dbc292a2a5 (matches checksum of downloaded file powerdown-1.02-noarch-unRAID.tgz)

Memory Usage: Light (10K to 500K)

Dependencies: none

[Flibblebot]

Sorry, should have mentioned that I'm using the custom powerdown script, not the unRAID poweroff script.

[Joe L.]

Are you sure?  Unraid also has a powerdown command.  Unless you give a explicit full path, you might be invoking the wrong command.

[graywolf]

might want to post a syslog here from the prior day.

 

if using the clean powerdown script, it should be copying the syslog somewhere on the flash drive) prior to shutdowns.

 

maybe something in there will help determine why unraid does not think it was shut down cleanly or why it didn't shut down cleanly

[Flibblebot]

Joe, I'm using the /sbin/powerdown script in my crontab.

Graywolf, the server hasn't been writing any logs to the flash drive, so I'm guessing that perhaps the unRAID server isn't being woken up properly by the other server before the power goes off.

 

Thanks for the help - just talking to someone else about this is helping my brain to start working like it should...

[graywolf]

1 other test....go thru manually what you have for your shutdown process (with unRaid awake).

Then bring it up how your have it set to do automatically and see if it does a Parity check also or not.

 

If it doesn't, then you are probably right, it isn't getting woken up properly to do your clean shutdown process and so when your powerstrip shuts off, an unclean shutdown occurred.

[Flibblebot]

OK, it's been running okay for the last few days, but I get up this morning to find that a parity check is running and the /boot/logs folder is unreadable.

 

I've had this happen before - I'm guessing that the flash drive hasn't finished writing the syslog files before the server shuts down. Would a "dirty" flash drive cause a parity check?

 

Does the powerdown script force a sync before shutting down?

[dgaschk]

Have you tested the wake-powerdown process while you watching the console? Run check disk on the flash in a PC.

[Flibblebot]

I've watched the powerdown process but haven't noticed anything untoward, and those times where the syslog was written OK, nothing seems out of place.

 

I have run chkdsk on the flash drive, but the only thing picked up is the buggered logs folder.

 

It just seems that sometimes the flash drive isn't quick enough to write the syslog before the shutdown completes.

 

Would a newer flash drive help?

[dgaschk]

The system won't shutdown until the write to the flash completes.

[Flibblebot]

So are the errors on the flash drive indicative that it's on its way out? The flash drive I'm using is a few years old, so would it be advisable to swap it out for a new one?

[dgaschk]

The flash is having problems from the unclean shutdowns.

[Flibblebot]

But without getting access to the syslog (that hasn't been written properly because of the unclean shutdown), it's not possible to know what's causing the errors, given that the syslogs do get written and a clean shutdown happens some of the time - thinking like that will trap your brain in an endless loop

 

Given that I'm fan of Occam and his shaving implements, surely the easiest thing would be to try a different flash drive and see whether the problem persists?

 

Alternatively, is there a way of stopping the powerdown script from writing a syslog?