Author Topic: Take backups! Really, take backups!  (Read 3724 times)

0 Members and 1 Guest are viewing this topic.

Offline Aaron Hulett

  • Administrator
  • Hero Member
  • *****
  • Posts: 1394
  • Schrödinger's cat walks into a bar... and doesn't.
    • View Profile
    • My Site
Take backups! Really, take backups!
« on: October 27, 2015, 04:09:34 AM »
It's been a while I suppose, but the home server has been up and running virtually 24/7 for the last 3 and a half years. So to check on things I had a look the hard disks' SMART statistics, and behold! 18 bad sectors were found (and uncorrectable) and another 18 were pending evaluation. Not exactly a good sign - it could mean failure is imminent or the drive could run for another year or two or ... so I opted to be safe and start replacing things. I was already planning to rebuild the box sometime next year so I sped that up and rebuilt it over the weekend.

CrystalDiskInfo is what I used (direct download for ZIP package without ads: http://crystalmark.info/redirect.php?product=CrystalDiskInfo) - run the executable (choose the one with 64 in the filename if you're on a 64-bit OS) and check your stats. If everything's green, should be good to go. If it calls out anything, may want to check into it further.

Regards,
Aaron

CrystalDiskInfo is what I personally used - there are other programs/products that will poll SMART details. I provided the name and link for convenience, and as you'd expect my thoughts and opinions here do not represent the thoughts or opinions of Microsoft nor does it create an endorsement. Microsoft is a registered trademark or trademark of Microsoft Corporation in the United States and/or other countries.

Offline Pierre75

  • Sr. Member
  • ****
  • Posts: 410
    • View Profile
Re: Take backups! Really, take backups!
« Reply #1 on: October 27, 2015, 05:03:08 AM »
Thanks for the heads up. :)  I run Ghost15 on both PCs. It does a Doc backup at 00.00 and a disk backup at 03.00 daily to separate attached USB hard drives. I have downloaded your link however. TU  ;)
IF IT AIN'T BROKE -  DON'T FIX IT

Offline JDBush61

  • Hero Member
  • *****
  • Posts: 4654
    • View Profile
Re: Take backups! Really, take backups!
« Reply #2 on: October 27, 2015, 06:00:37 AM »
Thanks for the heads-up, Aaron. My primary office computer is a 2010 Sony Vaio (Win 7 64-bit), and I sometimes wonder about the lifespan of the hard drive.

I downloaded CrystalDiskInfo (thanks for the link), did a check, and it gave me a yellow "Caution" for the health status. However, I have no idea why.  From ID 01 (Read Error Rate) to ID FO (Flying Head Hours) everything comes up green except 05 (Reallocated Sectors Count) and C5 (Current Pending Sector Count). The left-side buttons for those two are yellow (meaning caution, or something wrong I suppose).

The running temp is 47 degrees C, and shows as green. I have no idea what "Current", "Worst", "Threshold", and "Raw Values" refer to.

I did three screen captures (attached). Would greatly appreciate it if you could give them a look and tell me what you think.

Thanks!

"In an age when mass society has rendered obsolete the qualities of individual courage and independent thought, the oceans of the world still remain, vast and uncluttered, beautiful but unforgiving, awaiting those who will not submit. Their voyages are not an escape, but a fulfillment."

~ THE SLOCUM SOCIETY ~

Offline Digerati

  • LzD Friends
  • Hero Member
  • *****
  • Posts: 1037
    • View Profile
Re: Take backups! Really, take backups!
« Reply #3 on: October 27, 2015, 12:49:20 PM »
Your results seem odd JD, because if you follow the help screen, the color green is not even used as a status indictor. Blue means good, then yellow, red and gray.

It has been several years since I have used CrystalDiskInfo (since Speccy came out and reports S.M.A.R.T status). I think last time I tried it, it did not support SSDs, for some reason. I see now that it does, but oddly, in health status, it also reports a percentage under good. For my hard drive, it just says Good.
Bill (AFE7Ret)
Freedom is NOT Free!
2007 - 2018

Offline JDBush61

  • Hero Member
  • *****
  • Posts: 4654
    • View Profile
Re: Take backups! Really, take backups!
« Reply #4 on: October 27, 2015, 01:45:25 PM »
Thanks, Digerati. Yes, when I ran the CrystalDiskInfo program the results were originally shown in blue (and yellow... for 2 IDs). However, The program offers "change to green" (or something like that), which I did because Aaron mentioned "green".

Aaron: "If everything's green, should be good to go."

Anyway, the green buttons in my screen shots were originally blue. You are correct.
"In an age when mass society has rendered obsolete the qualities of individual courage and independent thought, the oceans of the world still remain, vast and uncluttered, beautiful but unforgiving, awaiting those who will not submit. Their voyages are not an escape, but a fulfillment."

~ THE SLOCUM SOCIETY ~

Offline Digerati

  • LzD Friends
  • Hero Member
  • *****
  • Posts: 1037
    • View Profile
Re: Take backups! Really, take backups!
« Reply #5 on: October 27, 2015, 03:20:32 PM »
Doh!  :-[ Yeah, didn't even see the Theme option.
Bill (AFE7Ret)
Freedom is NOT Free!
2007 - 2018

Offline Aaron Hulett

  • Administrator
  • Hero Member
  • *****
  • Posts: 1394
  • Schrödinger's cat walks into a bar... and doesn't.
    • View Profile
    • My Site
Re: Take backups! Really, take backups!
« Reply #6 on: October 29, 2015, 06:21:23 AM »
I did three screen captures (attached). Would greatly appreciate it if you could give them a look and tell me what you think.

Thanks!
You're in the "No need to panic but keep an eye on it/plan now" phase. There's one sector pending evaluation - it could be bad or it could be ok. The sector was either unreadable or the drive had a heck of a time reading it, so it's flagged so that the next time it's written to the drive will verify it's storing the data correctly. If it does store correctly, flag goes away and life goes on. If it doesn't, the sector is marked as bad and a backup sector is used in its place.

This is what happened with the one relocated sector. The drive relocated one sector over to the set of extras the drive has available. Not only do sectors start to go bad over time, the drive has bad sectors direct out of the factory (in other words, nothing's perfect). Back in the day with 8086 processors, so 25 years ago when having a 30 megabyte drive was HUGE, drives weren't as advanced and a label was attached to the drive with all the bad sectors. If you extremely wiped the drive, when you set it back up you had to manually enter all that info in so they wouldn't be used.

Today's drives are more self-managing at this, and hence these SMART trackers to monitor drive health. The idea is if an indicator suggests something's up, the system will alert you so you can take appropriate actions. Of course, if you are taking regular backups, this action will mainly be ordering a replacement drive and installing it. But if you didn't take a backup, this could be your last shot to grab one, or it might be too late. You're better off with a backup ahead of time, Captain Obvious said.

So with one relocated sector and one under evaluation, are you in a giant world of impending doom? It's going to depend on if this creeps up further. You definitely want to be taking backups in case this starts increasing, but with one bad and one in review it could be nothing out of the ordinary. The extra sectors that ship on the drive are there for this exact reason, meaning "the system worked". I'm of the camp that either you notice 1, and then 2, and then 4, and then... in a very short amount of time (weeks, maybe in a few months), OR you have more than 12 affected sectors (this would suggest to me something's up).

Now about the current versus worst versus threshold and raw data. Raw data is exactly that, the current data for that metric. So for reallocated sector count, you have 1. The "measure" of how awful this is comes from the current value. Each manufacturer can define what these measures are, but in this case it's at 100 and the worst it's ever been is 100. These count *DOWN* so higher is better. And threshold is when the drive's going to start saying, "Hey, pay attention to me!" and the system will give you a SMART alert. For the reallocated sector count, this'll happen when the current value passes the threshold.

And as you probably see, a lot of these have threshold of zero, meaning you'll NEVER get an alert on those values. Which is what I ran into. I had 18 uncorrectable sectors, which shows up at 12 in the raw value because these are in hex, and we count like this:

0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A, B, C, D, E, F, 10, 11, 12. 12 in hex (which is written as 0x12 or 12h) is 18.

I had another 18 pending sectors, but 18 *uncorrectable* sectors means potential data loss. And when I copied stuff off the drive, I had a few moments where copying failed, and one attempt that brought the system to a crawl but I got the data. This is because the drive was trying and trying and trying to get that data, and it'll try for a while, and while it's trying it's not doing anything else, so you wait for it to either be successful or give up.

It's this trying and trying that is horrid for things like network attached storage arrays or RAID arrays in a computer. These things require the drives to be up and running and responding, and if a drive stalls out like that the network attached storage device or the computer's RAID controller will flag the drive as potentially bad. They're expecting the drive to reply back VERY quickly. So for these uses, you'll see drives built for these uses, such as Western Digital Red drives. When I rebuilt this server I implemented a RAID array (RAID 10 for those familiar) using four 3TB Western Digital Red Pro drives. For these drives, if they hit one of these cases they'll try to recover for about 7 seconds and then either correct things or give up. If they correct things, the RAID array goes on its merry way; if not, the RAID controller flags puts the drive offline and raises an alert.

Why RAID 10? I wanted speed and redundancy. The RAID array's size is 6TB. Two of the 3TB drives are striped together (RAID 0) to create a 6TB storage volume (the operating system sees this as one 6TB storage area). Data is split across them, essentially equally, meaning a file is split in half and stored on each one. The other two drives mirror the first two (RAID 1) so that way if one drive fails I can replace it and the mirror image will be copied over to the replacement. Hence the name RAID 10 given both options are implemented.

But this is redundancy, not backup. So I still take backups of these using two removable hard drives. One stays in the machine, the other is stored offsite. They're swapped occasionally (weekly). This way if the box is hit by a power surge or destroyed in a catastrophe, I have my data offsite.

I'm considering cloud-based backups so that I can skip swapping drives around but I haven't found something that fits quite what I want. Yes, I store some of my files on OneDrive but I'd like to store the entire virtual machine (I implemented the server as a virtual machine) in the cloud so I can pull it down and turn it on. Not sure if Azure makes sense (backing up to Azure is an option in Windows Server Backup) but I'm looking into it.

So there you go, checking your drive's SMART values, understanding what they're trying to tell you, what to do if they've entered the "caution" stage, and what I did when mine entered the caution stage. Are you taking backups? If not, would you like to explore options and if so what edition of Windows 7 are you running (Home Basic, Home Premium, Ultimate, ...)?

Regards,
Aaron

This information is provided "AS IS" without warranty, and confers no rights. Opinions expressed are my own and do not reflect the views and opinions of Microsoft.

Offline Corrine

  • The Mystical Rose
  • Administrator
  • Hero Member
  • *****
  • Posts: 19463
  • "Stronger than the past, united in our goal."
    • View Profile
    • Security Garden
Re: Take backups! Really, take backups!
« Reply #7 on: October 29, 2015, 01:43:51 PM »
I'm considering cloud-based backups so that I can skip swapping drives around but I haven't found something that fits quite what I want. Yes, I store some of my files on OneDrive but I'd like to store the entire virtual machine (I implemented the server as a virtual machine) in the cloud so I can pull it down and turn it on. Not sure if Azure makes sense (backing up to Azure is an option in Windows Server Backup) but I'm looking into it.
I did some research of Microsoft options for someone with a question in another forum where the company the person worked for had suffered a loss of data and online transactions and his boss asked him to look into options for disaster recovery.  One of the articles I included was Configure Azure Backup Services to prepare for back up of Windows Server.  What I find impressive about Azure is the multiple storage replication.  There are three copies of data maintained on separate nodes with Local Storage.  A business in a geographic area that is prone to natural disasters could certainly benefit from the Geo-redundant storage which includes six copies of data.  Azure Storage replication.


Take a walk through the "Security Garden" -- Where Everything is Coming up Roses!

Remember - A day without laughter is a day wasted.
May the wind sing to you and the sun rise in your heart.

Offline Corrine

  • The Mystical Rose
  • Administrator
  • Hero Member
  • *****
  • Posts: 19463
  • "Stronger than the past, united in our goal."
    • View Profile
    • Security Garden
Re: Take backups! Really, take backups!
« Reply #8 on: October 29, 2015, 05:05:00 PM »
Coincidentally, this article was just published about a new service, Azure Resource Health: Reducing troubleshooting time with Azure Resource health.


Take a walk through the "Security Garden" -- Where Everything is Coming up Roses!

Remember - A day without laughter is a day wasted.
May the wind sing to you and the sun rise in your heart.

Offline JDBush61

  • Hero Member
  • *****
  • Posts: 4654
    • View Profile
Re: Take backups! Really, take backups!
« Reply #9 on: October 30, 2015, 03:43:41 AM »
I did three screen captures (attached). Would greatly appreciate it if you could give them a look and tell me what you think.

Thanks!
You're in the "No need to panic but keep an eye on it/plan now" phase. There's one sector pending evaluation - it could be bad or it could be ok....

Aaron, sincere thanks for your long and detailed response! Much of what you posted went over my head, so I'll do some further investigation on my end in order to better understand the details. I do fairly regular backups on this machine, yet will pay even closer attention from this point forward.

Thanks again!

John
"In an age when mass society has rendered obsolete the qualities of individual courage and independent thought, the oceans of the world still remain, vast and uncluttered, beautiful but unforgiving, awaiting those who will not submit. Their voyages are not an escape, but a fulfillment."

~ THE SLOCUM SOCIETY ~