I started preclearing my new Western Digital WD20EURS 2TB drive yesterday afternoon. It's been 16 hours and 31 minutes and it's 1% done with step 2 of 10 (copying zeros to remainder of disk to clear it).

 

Just like my last preclear session IT IS VERY SLOW. It is only copying at 2.5 MB/s.

 

I assumed it was slow last time because of a defective drive. However, that drive was still passing as healthy according to its SMART report. Now this brand new drive is preclearing very slowly. I wonder what's going on here.

 

Is this slow preclear speed because I have an array that is stopped, has an invalid configuration and reporting an error with too many missing disks?

My thoughts are:

1. Preclear new drives thoroughly, preferably installed in the physical slot they will occupy permanently.

2. While the preclearing is proceeding, you could set a new configuration using only the good drives (parity, 1, 3, 5, 6)

3. After the initial calculation of parity is done, kick off a non-correcting parity check.

4. Once step 3 is done, browse around the array and see if your remaining data looks good.

5. Add the drives you precleared in step 1 to their array slots, one at a time, and kick off another non-correcting check after each new drive is added.

6. After the array is happy with both new drives, copy your backed up data from the old 2 and 4 back into place.

7. Return disk 2 after you are satisfied that any data from disk 2 is back in place.

8. Ponder a backup strategy that doesn't use this specific unraid server. A second unraid tower in another location, USB drives, whatever.

 

Step 1 and 2 will likely take several days.

 

How do I set a new configuration with the good drives? The unRAID main menu won't let me pick 'START', it is grayed out. Next to the start button it says "Too many wrong and/or missing disks." In the disk status area all my good drives are flashing green, the disk2 position is dark red (not installed), and the disk4 position is red (Missing). In the Devices tab, disk4 is assigned, which is probably why unRAID is reporting it as missing. If I unassign it, it doesn't change anything in the main menu, disk4 still missing.

 

Should I toggle to a new screen and type "initconfig" in my telnet session (this is where I'm monitoring the pre-clear)? Or, I suppose I could just type "initconfig" at the command line of the system console. I have a keyboard and monitor connected at the server.  With everything that's happened recently, I am very cautious about doing anything that might make matter worse.

 

 

No, your array status has no effect on the preclear. You have some kind of hardware issue it appears.

 

And yes, use initconfig to reset the array. Assign the disks you want in the array and initconfig to get a valid array so you can start.

 

 

No, your array status has no effect on the preclear. You have some kind of hardware issue it appears.

 

And yes, use initconfig to reset the array. Assign the disks you want in the array and initconfig to get a valid array so you can start.

 

How do I diagnose my hardware issues? I've run memtest and haven't produced any errors; I've replaced my power supply; what next?

 

Could you be more specific about using initconfig?

 

Do I type "initconfig" at the command prompt and then respond "Yes"

 

Then I go back to the Media - Main screen in my web browser and START my array...

 

Is that correct?

 

Thanks for your help.

I'm not sure on the hardware. Try a different port first.

 

As for the initconfig - yes that appears correct.

OK, I've run initconfig and started the array. Parity-Sync is now in progress.

 

BTW, the pre-clear has steadily sped up, not a lot, but it is improving. The current speed is 6.5 MB/sec and I'm at 13% of step 2.

The new configuration is set. Parity is valid. I've run SMART status reports on all of the remaining good drives and there are no pending sectors or anything.

 

The new drive I purchased during this crisis is still preclearing. After 40 hours and 13 minutes, it is still at 32% of step 2. It is running at a rate of 6.6 MB/sec.  I'm tempted to cancel the preclear and try it from one of my different SATA controllers on my motherboard. I could also try a different SATA cable.

 

Would there be any risk to this drive if I were to cancel the preclear? I could always wait for another 80 hours for it to finish.

Cancel it. It should be running well over 10x that speed.

What will I accomplish by canceling? Are you suggesting I cancel and then change my SATA cable and/or switch to another port and then try running preclear again?

my server doesnt play well with 16gb flash drives either. it actually refuses to boot with anything larger than 4gb.

 

i would cancel that pre-clear, its not running right. plug the drive into a different sata port and try again. if the results are the same change the cable.

my server doesnt play well with 16gb flash drives either. it actually refuses to boot with anything larger than 4gb.

 

i would cancel that pre-clear, its not running right. plug the drive into a different sata port and try again. if the results are the same change the cable.

 

I was wondering if there was any relationship between my 16 Gb flash drive and the problems I'm having here. I couldn't run memtest, but it could run unRAID. So, before swapping ports and/or cables I decided to try rebooting my server with my old 128Mb flash drive, which still has my Plus license. At this point, I'd have to use initconfig and create a new configuration to work with this license, so I don't think I should mess with that.

 

Now I'm running the preclear off of this Plus license and it's showing what appears to be normal performance - 57 MB/sec.

 

So could all these problems be related to my 16Gb flash drive?

 

If it is, it's a relief to finally have an answer, but it's also really, really frustrating to realize that I just purchased a new 2TB hard drive, a power supply and shipped off what might be 2 perfectly good hard drives, with there associated shipping costs. On top of that, I couldn't get approximately 64 Blu-rays and/or DVDs backed up. So they need to be re-ripped. Aaarrrrgggg. 

One other thing

I now have a monitor connected to my server, so I can look at the screen during boot-up. When it finishes loading the unRAID OS, the last few lines are:

 

Starting Samba: /usr/sbin/nmbd -D

                          /usr/sbin/smbd -D

/var/tmp/go: line 11: syntax error: unexpected end of file

 

 

Is this error anything to be concerned with?

 

BTW, the preclear speed is fluctuating around 60 MB/sec. That's a pretty normal speed, right?

I may have posted too quickly. The preclear is 31% into the pre-read and has slowed down to 40 MB/sec. I hope this isn't a bad sign.

This is very frustrating. The preclear is now on step 2 of 10 and it is at 8%. The speed at this time is 11 MB/sec.

 

Should I stop and try again with a different cable or connect to a different port? Or, should I just let it keep going?

 

This is very frustrating. The preclear is now on step 2 of 10 and it is at 8%. The speed at this time is 11 MB/sec.

 

Should I stop and try again with a different cable or connect to a different port? Or, should I just let it keep going?

 

Don't stop quite yet, grab a syslog first and post it here, zipped if necessary.  If the slowness is caused by disk errors or cable issues, then the syslog will show that.

 

One other thing

I now have a monitor connected to my server, so I can look at the screen during boot-up. When it finishes loading the unRAID OS, the last few lines are:

 

Starting Samba: /usr/sbin/nmbd -D

                          /usr/sbin/smbd -D

/var/tmp/go: line 11: syntax error: unexpected end of file

 

 

Is this error anything to be concerned with?

 

Can up also add a copy of your go file to the zip?

Here's the syslog and go file, which I just renamed to a txt file. This go file is from my old flash drive that was used for my Plus version of unRAID and is now being used to run the preclear.

syslog-2013-02-24.txt

go_file.txt

The go file was missing a line feed after the 'done', possibly causing a bad end of loop plus keeping UnMenu from loading.  It also has an extra space at the end of the 'install packages' line 11, which may have caused it to look for another parameter, but found the end-of-file.  I'm not sure you wanted to load any packages, so I commented the line out.  I've attached a fixed version of your go file (you'll need to rename it).

 

This 126GB flash drive (!) has UnRAID v4.7 installed, which defaulted to creating disks that start on sector 63 (I believe).  Did you change that default to start on 64, or did you make sure to use Preclear with the -A option for 4k alignment?  If not, that might be one reason for slower performance.  I'd recommend upgrading this flash drive to RC11.

 

There are NO disk errors, including cable errors in this syslog, so your cables are probably fine, and it's not a 'disk error' issue.  The drive was setup without issue, and at full speed (SATA 3.0Gbps and UDMA/133), and within this syslog there are no slowdowns issued to the drive.

 

It looks like you started the Preclear at 4:42pm, no partition table found on drive.  Nothing important happens that is logged after that, and the syslog last entry is at 4:28am the next morning, so there's nothing left to learn here.  Sorry.

go_GreggP.txt

The reason I'm using the 128MB flash drive is because I thought I was having problems with a 16GB flash drive.

It's a long story... I will summarize here:

 

1. Last fall, I had a power outage, which resulted in lots of errors when unRAID automatically ran a parity check, after I restarted the server. A syslog at the time revealed that I had lots of pending sectors and it was recommended that I replace the disk that was in the disk2 position. I immediately purchased a new drive and RMA'd the defective drive. Since I was increasing my capacity once the defective drive's replacement was installed, I needed to upgrade from v4.7 Plus to Pro. My 128MB flash drive was nearly full and I wanted to make sure I had enough space for any future releases and packages. So, when I got my Pro key, I had Tom give me a new key for the 16GB flash drive. This flash drive was one that I had laying around and wasn't being used for much, so I thought it would be a good upgrade/replacement for the 128MB drive, which I think I'd had for a very long time. Main point - I switched from Plus with a 128MB flash drive to Pro with a 16GB flash drive. This setup ran without issue until this current problem (see 2).

 

2. Near the end of January (at the start of this thread), I noticed extreme performance problems with writing to my unRAID server. The syslog and subsequent SMART test showed I had a drive problem again. The new drive in the disk2 position had pending sectors. This time, it was taken offline and precleared. After the preclear, the SMART test on the drive showed it no longer had any pending sectors. So the drive was then reassigned to the array. This also solved the performance problems. Writing to the server was back to normal. Until a week or so later.

 

3. The performance problems came back. I ran SMART tests on all my drives and they revealed that 2 drives had pending sectors - disk2 and disk4. Based on my previous experience, I went ahead and took disk2 offline and started preclearing. Which might have been a mistake. After it completed the preclear (in the normal amount of time) disk2 still had pending sectors. Since disk4 was still mounted, I still had a working array. I did my best to start backing up everything from these drives to other computers on my network, but I didn't have enough space to back up everything. Some responses to this thread suggested replacing my power supply because a failing power supply might cause these sort of problems. I decided to buy both a new 2TB drive and a new power supply.

 

4. Then I had to go out of town for a couple days. Upon returning, my server was running, but unRAID wasn't running. So I rebooted the machine. The system startup info showed errors with my port04. This error only appeared when disk4 was connected to port04. If I connected a different drive to that port the error disappeared. If I connected disk4 to a different port, the error appeared on that port. So now I had 2 drives that couldn't be assigned to the array.

 

5. By this time my new 2TB drive and power supply had arrived. So I decided to go ahead and replace the power supply and remove disk2 and disk4. I installed the new 2TB drive in the disk2 slot. I have since been trying to get it precleared and it has gone extremely slowly. Which seems to indicate an underlying hardware problem. I was told to run memtest, but it wouldn't run and I discovered my keyboard was defective. Replaced the keyboard and memtest still wouldn't run. Tried installing a new version of memtest on my flash drive and it still wouldn't run. Finally, tried running memtest from my old 128MB flash drive and it work fine. The test showed there was nothing wrong with my RAM. I tried preclearing again, using the unRAID running from my 16GB flash drive, and it was very slow again - speeds of 1 - 2.5 MB/sec. People told me there was still something wrong with my hardware and to stop the preclear and maybe try a different SATA port and/or cable. Then someone responded to this thread that they couldn't even get their server to boot with anything larger than a 4GB flash drive. So I decided to restart unRAID with the 128MB flash drive running my old Plus version of 4.7 and then try the preclear. I wanted to try this test before switching to a different port or changing the cable. Running preclear with the 128MB flash drive seemed to work fine. During the pre-clear-read, it was running at around 60 MB/sec, which I assumed was the normal speed. It slowed down to as low as 30 MB/sec, but before I went to bed last night, it was up above 50 MB/sec. Then this morning it was at step 2 - zeroing data and the speed was slower than 12MB/sec. I assume this is well below the normal speed, but still a lot faster than it was when I was using my 16GB flash drive. I've ordered a 4GB flash drive from newegg and plan to ask Tom to switch my key to it after it arrives.

 

So, I'm still trying to determine if I have any other underlying hardware issues.

 

I'm usually very cautious about running beta or pre-release software on working systems. Is v5 RC11 as stable as or more stable than v4.7? If so, that's what I'll do when I switch to the new 4GB flash drive. However, I'm concerned that there is still some undiscovered issue with my hardware.

 

Thanks for looking at my syslog and go file and for editing the go file. I'll give it a try after this preclear finishes. By then I may even have a new USB flash drive.

 

I almost forgot... Yes, I ran preclear with the -A option because the drive, a Western Digital WD20EURS, is an advanced format drive.

I'm tempted to buy a new motherboard, CPU and RAM, because I just don't know what else I can do here. Anyone want to buy an ABIT AB9 Pro with CPU, RAM and video card?

 

Also, I'd really like some feedback on the stability of unRAID v5 RC11.

 

Thanks

 

BTW, the preclear on this drive has slowed down to 8 MB/sec and I'm only at 46% of zeroing (step 2) and it has been going for over 41 hours.

I just checked the progress of my pre-clear and it looks like there are lots of errors at the end of my syslog. For the past hour or so, it is showing:

 

Media kernel: lost page write due to I/O error on sde

Media kernel: Buffer I/O error on  device sde, logical block ...

 

 

I'm preclearing sde, which is the drive listing these errors.

 

Does that mean I have a bad drive, connection, port?

Should I cancel the preclear?

 

I'd appreciate any help I can get on this.

 

Thanks

syslog-2013-02-25.txt

Sorry, my extremely low paying job gets in the way of helping here.  The syslog you attached has an additional 34 lines of the same session.  What is very strange is that the "Unhandled error code" and "end_request: I/O error" statements, and the series of "Buffer I/O" and "lost page write" errors all occur without a SINGLE exception being raised!  The exception handler is not called once here, and normally those error statements follow a series of lower level exception handler lines.  Do you know of anything that might be filtering the syslog you have attached?

 

I would cancel the Preclear, obviously something is seriously wrong, and a 'cleared' drive with failed writes is not useful.  I would still recommend either upgrading this flash drive to RC11 or using another flash drive with RC11 and Preclear on it, and trying again.  There should be better error reporting, if nothing else.  Sorry this is all taking a very long time.

 

Does that mean I have a bad drive, connection, port?

Sorry, cannot make any conclusions at all yet.

I'm sorry, I didn't mean to be impatient with you. I'm probably getting a lot more stressed out about this than I should. After all, it's just a hobby. Potentially more expensive than I hoped, but never the less, just a hobby.

 

I have no idea if anything is filtering my syslog. That syslog was generated with the syslog utility that's part of unMenu. This attached syslog was one I created at the command line. It's identical, except the last 3 lines.

 

On the positive side, the replacement drive for my RMA'd Western Digital 2 TB drive arrived today. The weird thing is they gave me a 3TB drive as a replacement for my 2 TB drive. Unfortunately, it doesn't do me any good at this time. It is now the only 3 TB drive I own, and since my parity drive is only 2 TB, that's all I can take advantage of in my array. Other than using it as my new parity drive, can anyone thing of other ways I can take advantage of the extra terabyte?

syslog-2013-02-25_2.txt

At any other time and with a healthy array, a new 3TB drive would have been a nice upgrade for your array, but not now.  It's just one more unnecessary complication.  You can't catch a break, can you...

I cancelled the preclear. Then I downloaded v5 RC11 and unzipped the contents to my PC. I then shut down the unRAID server, extracted the thumb drive and plugged it into my PC. Then I copied the v5RC11 bzimage and bzroot files to my thumb drive and replaced the old versions of those files. I also renamed your edited 'go (GreggP).txt' to 'go' and replaced the one in my config folder.

 

After rebooting my server, it hangs on bzroot. I get 7 lines of dots and then the cursor stops about 2/3rds across the screen.

I thought that might happen, I should have mentioned the possible need to run make_bootable.bat if it did not boot, since this was an older flash drive.  Please see Migrating from unRAID 4.7 to unRAID 5.0, the special instructions in the 'Reboot and test' section.