Recently upgraded to 4.7 from 4.6 in preparation for the step to version 5. I was writing a file out to the server and I noticed 200 errors showing in the webgui under the parity drive. I've included a log and a smart readout of that drive.

 

I continued to write other files out to the server after the first, WITHOUT  further errors. The movie looks okay, but I'm watching it tonight to make certain.

 

Can someone tell me what they reckon. My server has had very little issue over the many years I've been running it. Always makes me concerned when an upgrade has taken place in the last couple of days.

 

Are there any steps I should take?

 

I plan another parity check across Thursday night so I can upgrade to version5 on the weekend.

 

Appreciate the help.  Cheers

syslog-2013-09-18.txt

ParityDriveSmartInfo.txt

The SMART report looks okay.

 

It's likely you simply had a glitch in updating parity during the writes.

 

Run a correcting check ... then run another one after it.    The second one should have zero sync errors. ... as long as that's the case, you're in good shape and ready to upgrade.

 

The SMART report looks okay.

 

It's likely you simply had a glitch in updating parity during the writes.

 

Run a correcting check ... then run another one after it.    The second one should have zero sync errors. ... as long as that's the case, you're in good shape and ready to upgrade.

 

Running a parity check now. It takes 22 hours, so I won't know until tonight. So far, still no sync errors have shown up from the parity check.

Recently upgraded to 4.7 from 4.6 in preparation for the step to version 5. I was writing a file out to the server and I noticed 200 errors showing in the webgui under the parity drive. I've included a log and a smart readout of that drive.

There's a known bug in v4 that can cause data corruption if you write to the server while parity check is running.  I can't imagine a reason why you don't upgrade straight to v5.

 

I've never written to the server during a parity check. I upgraded to 4.7 because that's what the upgrade instructions say to do before upgrading to V5. Hopefully, everything will go perfectly and I can finally make the switch to >2Tb drives.

 

Appreciate the comments. Thx.

Recently upgraded to 4.7 from 4.6 in preparation for the step to version 5. I was writing a file out to the server and I noticed 200 errors showing in the webgui under the parity drive. I've included a log and a smart readout of that drive.

There's a known bug in v4 that can cause data corruption if you write to the server while parity check is running.  I can't imagine a reason why you don't upgrade straight to v5.

 

I've never written to the server during a parity check. I upgraded to 4.7 because that's what the upgrade instructions say to do before upgrading to V5. Hopefully, everything will go perfectly and I can finally make the switch to >2Tb drives.

 

Appreciate the comments. Thx.

 

This step is only required if you want to keep configs. A clean install may be faster.

Okay...my parity check has finished and it said NO errors. BUT...at the end of the webgui line for the parity drive, it says 1 write and 19 errors.  So what is going on?? The log shows the same type of errors as the one supplied. In this case, nothing has been written to the server. It has only been doing a parity check. I'm somewhat concerned. Could it be something to do with the 4.7 upgrade. Until I did this, I had NO problems whatsoever.

 

Here is the new log.

 

Do I need to do another parity check? Is there something else I should do? Thx

 

Edit: I found this post http://lime-technology.com/forum/index.php?topic=25802.msg224962#msg224962 that mentions the Current_Pending_Sector RAW_VALUEs must always be zero. I notice mine is now 8, whereas before the parity check it was 3. Neither is zero. The post states the parity disk needs to be rebuilt. Is this the same as a parity check or something else?

syslog-2013-09-19.txt

The parity drive needs to be rebuilt. A correcting parity check may work. A better method is to start the array with parity unassigned and then reassign parity and start the array. You can pre-clear the drive before reassigning for a more rigorous repair. If you have a pre-cleared spare use it now and then pre-clear the current disk. If it's pending sectors go to zero then use it as the spare.

(a)  You need to rebuild parity to force writes on all sectors on that drive -- this will reallocate all of the pending reallocations.

 

(b)  Since you have to do this, you may as well simply start from scratch with v5.0 => save your current flash drive contents (especially your key file);  reformat it and put v5.0 on it -- add your key file;  boot to it; assign your data drives; Start the array to confirm it looks okay;  Stop the array; assign your NEW parity drive;  then Start the array and let it do a parity sync.

 

©  Now do a pre-clear on your old parity drive to give it a thorough checkout;  then you can add it to the array.

 

Thx muchly for the help. I was doing another parity check overnight, but there was a power failure and the server shut down, so it didn't complete.

 

Upgrading to V5 sounds like the way to go...BUT...am I currently protected?? OR, is my parity suspect because of the errors. The upgrade instructions say to make sure parity is correct. Can I trust it?

 

I was going to do a fresh install, but copy my configs folder across. If I don't copy the configs folder across - does unRAID allow me to assign the data drives into the correct order before it starts (meaning-does it start in the array stopped position?

 

I've got concerns on so many levels

 

EDIT: Ive upgraded to V5 and everything seemed to go well. I had a nervous moment when I couldn't access the gui after the power outage last night (even though I'm on an UPS??). I had to power-off the router, network switch and server for it to gain access again.

 

I am currently running the parity update.

 

A couple of things I did notice...

 

1 - The linux designation of the drives was different (sda etc was something else)

 

2 - Default partition format says MBR: 4k-aligned in the DIsksettings page, while the individual disks say MBR: Unaligned. What is this? Does this need to be changed?

 

Thx muchly for the help. I was doing another parity check overnight, but there was a power failure and the server shut down, so it didn't complete.

 

Upgrading to V5 sounds like the way to go...BUT...am I currently protected?? OR, is my parity suspect because of the errors. The upgrade instructions say to make sure parity is correct. Can I trust it?

 

I was going to do a fresh install, but copy my configs folder across. If I don't copy the configs folder across - does unRAID allow me to assign the data drives into the correct order before it starts (meaning-does it start in the array stopped position?

 

I've got concerns on so many levels

 

EDIT: Ive upgraded to V5 and everything seemed to go well. I had a nervous moment when I couldn't access the gui after the power outage last night (even though I'm on an UPS??). I had to power-off the router, network switch and server for it to gain access again.

Are you running apcupsd?

I am currently running the parity update.

 

A couple of things I did notice...

 

1 - The linux designation of the drives was different (sda etc was something else)

normal.

 

2 - Default partition format says MBR: 4k-aligned in the DIsksettings page, while the individual disks say MBR: Unaligned. What is this? Does this need to be changed?

 

 

This is normal. All new disks will be 4k-aligned.

The upgrade seemed to go well. I carried out a fresh install using my exisiting configs folder. Worked great. I also re-installed unmenu, apcupsd, powerdown and cache_dirs. All going great.

 

It'll be interesting to see how long the parity check takes under V5. It was about 22 hours under 4.7.

 

What does changing the security on the flash do - Private vs Secure. I guess one or both will required a password?

 

Again...the help is much appreciated.

Okay...the parity check completed without error.

 

But...the "Current_Pending_Sector" still shows @8. So what is going on. unRAID is reporting that parity is valid - completed without error - but I saw it stated that Current_Pending_Sector greater than zero was an issue.

 

Is this a bug in unRAID. (also, when the server boots, all the drives show a "PASSED" flag). What should I do?

 

Here is the latest smart report.

ParityDriveSmartInfo-V5_After_ParityCheck_report.txt

Does the parity drive show any errors on unRAID main? Rebuild parity by un-assigning parity and starting the array then stop the array and assign parity. Start again and parity will rebuild. The pending sector count should be zero.

Does the parity drive show any errors on unRAID main? Rebuild parity by un-assigning parity and starting the array then stop the array and assign parity. Start again and parity will rebuild. The pending sector count should be zero.

 

NO - the unRAID main page shows zero errors. It also says that parity is valid and completed the check with zero errors. All the drives (at the end of each line) also shows zero errors.

 

From this, I have a couple of questions...

 

1 - unRAID shows this as perfectly safe (is it?). If the drive hadn't showed any errors, I wouldn't have checked the smart info. So unRAID obviously still thinks everything is okay, but if the Current_Pending_Sector isn't zero, then this appears to be an issue.

 

2 - Unassigning the parity drive will kill the parity protection will it not?? So if I suffer a drive failure then I will lose all data?

 

I struggle to understand how unRAID can report everything fine if it's not! If the Current_Pending_Sector isn't zero, then shouldn't a flag be raised within unRAID to say something is amiss?

 

I'm thinking of upgrading my parity drive to a larger one anyway, but will have to order one in.

 

Thx for your help.

 

http://en.wikipedia.org/wiki/S.M.A.R.T.

http://en.wikipedia.org/wiki/S.M.A.R.T.#ATA_S.M.A.R.T._attributes

 

Current Pending Sector is the # of sectors the disks firmware of the drive. While they've probably read back correctly 99 out of 100 times, there has been at least one failure on that sector. This could be by a re-read of the sectors with a cheap md5 not matching, or other similar issues. Basically SMART is saying 8 sectors on your harddrive can't be trusted 100%.

 

This while sounding bad is *NOT* the end of the disk. It just means that what it wrote and read back weren't the same. On modern Hard disks they expect this over the life of the drive.

 

HOWEVER, clearing them isn't exactly straight-forward. I've found removing the disk and doing 3 or more preclears on it will force the bios to reallocate the pending sectors, or clear them as good. The downside? Doing this on a parity drive, or other drive in the array will leave it in an unprotected state.

 

Safest method? Buy new drive of equal size, replace it with possible failing drive. Rebuild parity. After parity has been rebuilt, run a pre-clear on that drive with at least 3+ cycles. If it shows as 0 pending after the pre-clears, you've got them reallocated or reassigned as good.

 

If it doesn't? Well..... what's 200.00?

 

Okay...the parity check completed without error.

 

But...the "Current_Pending_Sector" still shows @8. So what is going on. unRAID is reporting that parity is valid - completed without error - but I saw it stated that Current_Pending_Sector greater than zero was an issue.

 

Is this a bug in unRAID. (also, when the server boots, all the drives show a "PASSED" flag). What should I do?

 

Here is the latest smart report.

That helped a lot . As I have a spare 2Tb drive here, I'll stick that in and rebuild parity. I guess this will leave me unprotected until the rebuild is done, but I can use the existing parity as a fall-back - especially as unRAID says it's okay.  Thx again.

Does the parity drive show any errors on unRAID main? Rebuild parity by un-assigning parity and starting the array then stop the array and assign parity. Start again and parity will rebuild. The pending sector count should be zero.

 

NO - the unRAID main page shows zero errors. It also says that parity is valid and completed the check with zero errors. All the drives (at the end of each line) also shows zero errors.

 

From this, I have a couple of questions...

 

1 - unRAID shows this as perfectly safe (is it?). If the drive hadn't showed any errors, I wouldn't have checked the smart info. So unRAID obviously still thinks everything is okay, but if the Current_Pending_Sector isn't zero, then this appears to be an issue.

 

2 - Unassigning the parity drive will kill the parity protection will it not?? So if I suffer a drive failure then I will lose all data?

 

I struggle to understand how unRAID can report everything fine if it's not! If the Current_Pending_Sector isn't zero, then shouldn't a flag be raised within unRAID to say something is amiss?

 

I'm thinking of upgrading my parity drive to a larger one anyway, but will have to order one in.

 

Thx for your help.

 

Have you restarted since the  parity check?

http://en.wikipedia.org/wiki/S.M.A.R.T.

http://en.wikipedia.org/wiki/S.M.A.R.T.#ATA_S.M.A.R.T._attributes

 

Current Pending Sector is the # of sectors the disks firmware of the drive. While they've probably read back correctly 99 out of 100 times, there has been at least one failure on that sector. This could be by a re-read of the sectors with a cheap md5 not matching, or other similar issues. Basically SMART is saying 8 sectors on your harddrive can't be trusted 100%.

 

This while sounding bad is *NOT* the end of the disk. It just means that what it wrote and read back weren't the same. On modern Hard disks they expect this over the life of the drive.

 

HOWEVER, clearing them isn't exactly straight-forward. I've found removing the disk and doing 3 or more preclears on it will force the bios to reallocate the pending sectors, or clear them as good. The downside? Doing this on a parity drive, or other drive in the array will leave it in an unprotected state.

 

Safest method? Buy new drive of equal size, replace it with possible failing drive. Rebuild parity. After parity has been rebuilt, run a pre-clear on that drive with at least 3+ cycles. If it shows as 0 pending after the pre-clears, you've got them reallocated or reassigned as good.

 

If it doesn't? Well..... what's 200.00?

 

Okay...the parity check completed without error.

 

But...the "Current_Pending_Sector" still shows @8. So what is going on. unRAID is reporting that parity is valid - completed without error - but I saw it stated that Current_Pending_Sector greater than zero was an issue.

 

Is this a bug in unRAID. (also, when the server boots, all the drives show a "PASSED" flag). What should I do?

 

Here is the latest smart report.

 

If there are pending sectors after a single pre-clear the disk is bad.

You can only use your old parity disk as a fall-back only if: (1) backup the superblock file from the flash disk after stopping the arry, AND (2) don't mount any data disks after removing the old parity disk. (disable autostart, start in maintenance mode.).  If you make the mistake of mounting any data disks, then something will be written to those disks, and that will make the old parity disk useless as a fall-back in case of a disk failure before the new parity disk is built.

 

Sounds tricky.  If I start in maintenance mode, I won't have my UPS and the apcupsd stuff happening. Booting into a stopped array will leave the data disks unmounted won't it.

 

If there are pending sectors after a single pre-clear the disk is bad.

 

I'm going to replace the parity with a new drive and take the risk nothing will break while the array is unprotected. I have a couple of new 2Tb drives spare. I'll stick to a 2Tb as I don't know if my Asus P5B-E motherboard can handle drives greater than 2Tb.

Thx

Sounds tricky... I won't have my UPS and the apcupsd stuff...

You'll still be physically plugged into the UPS, right?  Why would apcupsd matter?

 

If there is an extended power outage you'll have to shutdown manually.

Well, the parity rebuild took place without issue...and I've since done a parity check to verify it - also without issue. Excellent.

 

I mistakenly assumed the comment about "Maintenance Mode" was referring to "Safe Mode" on boot. Being new to V5 I'd never seen it before, but it popped up during the parity rebuild. So now I know what to expect for next time.

 

Can I ask my question about the motherboard again, regarding the maximum sized HDD it can take.  (it's a Asus P5B-E (NOT the plus version)) Line here - http://www.asus.com/Motherboards/P5BE/

 

I haven't found a lot of info regarding this aspect. Doesn't help that I don't really know what terms to search for. I have found this post http://vip.asus.com/forum/view.aspx?id=20070101183532054&board_id=1&model=P5B+Deluxe%2FWiFi-AP&SLanguage=en-us&page=1 that states a lot of technical info and seems to say that some of the controllers do in fact have the ability to have drives larger the 2Tb.

 

Is there a command that will tell me? Don't want to buy a large drive to find it won't all be available.  Appreciate the help.

I've seen mixed reports on whether or not a P965 chipset supports > 2TB drives, but it seems most likely that it does not.    However, the JMicron JMB363 controller does support them ... so it won't hurt to buy one 3 or 4TB drive and try it.    Be sure to update the Intel RST drivers -- IF the drives will work on the chipset controller, you'll definitely need the latest RST drivers.    If the drive won't work on your chipset controller, you can still use it on the JMicron ports.

 

I've seen mixed reports on whether or not a P965 chipset supports > 2TB drives, but it seems most likely that it does not.    However, the JMicron JMB363 controller does support them ... so it won't hurt to buy one 3 or 4TB drive and try it.    Be sure to update the Intel RST drivers -- IF the drives will work on the chipset controller, you'll definitely need the latest RST drivers.    If the drive won't work on your chipset controller, you can still use it on the JMicron ports.

 

So are you saying that only select ports will run a >2Tb drive? I think the JMicron has two, but I don;'t think the parity drive is on it. That would mean I'd have to swap drive positions seeing the Parity has to be one of the largest drives.

 

How do you update the MB drivers under linux? Is this something done in the bios? The Intel RST driver page only really allows you to select windows O/S's.

 

Looks like it'll probably be easier and safer to just stick with 2Tb drives.