Smartmon error message in the installer

[BitJam]

I think we should address the question of what to do if the smartmon tools say the disk passes while our test says it fails. I imagine there is near universal agreement that if the smartmon tools say the disk is failing then we should suggest they abort the install. The answer to the question of what to do when the tests give different results will guide what we do in the installer.

For example, if we can get that pass/fail bit of information via a script or program then I suggest we offer two different warnings. If it is only our raw count test that fails then we just say the failure rate will be higher than average. If smartmon says the disk is failing then we suggest they abort the install and use a different physical disk. I think this was a point of confusion. Some users didn't understand that the smartmon errors will be the same on all partitions of a physical drive. We don't want people to waste time selecting a different partition on the same drive only to find they get exactly the same error message.

I still encourage other people to investigate and not rely solely on my interpretation of the Google paper.

[Utopia]

Is this OK?

To me it looks perfect. Don't know how other people react to the message, but it feels like it shouldn't scare anyone from installing MX.
Henry

[jdmeaux1952]

It is the error message that is causing the panic. As I was installing MX-14.1.1 on my HD, I too received that error. I did stop, but then I used my "other tools" to check on my HD. (I forgot how clunky DOS and Pascal was.) According to my findings, my entire HD was is above average condition. Less than 5% bad sectors over the entire drive (already blocked). Less than 30% of estimated maximum usage time. No need to panic.

It reminded me of the children of the 1960s and 70s, who did not meet the criteria that Dr. Spock had set as being "normal" for a child. How many mothers ran to the doctors because their child didn't meet that criteria? And how many of these children turned out to be ABOVE NORMAL?

In the famous words of Bob McFerrin, "Don't worry. Be happy."

[Adrian]

I think we should address the question of what to do if the smartmon tools say the disk passes while our test says it fails. I imagine there is near universal agreement that if the smartmon tools say the disk is failing then we should suggest they abort the install. The answer to the question of what to do when the tests give different results will guide what we do in the installer.

For example, if we can get that pass/fail bit of information via a script or program then I suggest we offer two different warnings. If it is only our raw count test that fails then we just say the failure rate will be higher than average. If smartmon says the disk is failing then we suggest they abort the install and use a different physical disk. I think this was a point of confusion. Some users didn't understand that the smartmon errors will be the same on all partitions of a physical drive. We don't want people to waste time selecting a different partition on the same drive only to find they get exactly the same error message.

I still encourage other people to investigate and not rely solely on my interpretation of the Google paper.

That's exactly what the code in the installer does. It checks first for FAILED. if Smartctl reports failed then it gives a stern message:

Code: Select all

The disk you selected for installation is failing.
For more information run "smartctl -A <drivename>" in console, as root.
You are strongly advised to abort. 
Do you want to abort the installation?"

We still let people continue, if they want to ignore the message.

If the smartmon doesn't report FAILED, but if some important attributes: 5, or 196, or 197, or 198 show counts > 0 then it gives this message:

Code: Select all

"Smartmon tool output:

<output> 

The smartmon tools report that the disk you selected for installation might have a higher than average failure rate in the upcoming year.

Do you want to continue?"

Are you OK with this?

[BitJam]

I think this looks much better. I want to think about it a bit. There may be a slight confusion from "the disk you selected for installation is failing" because people select a partition not a disk. This will lead some people to select a different partition on the same disk.

When people got the previous raw count warning it suggested they run smartmon manually which implied that the program had not already checked to see if smartmon reported pass or fail. We might want to make it known that the test was run and the disk passed:

The disk with the partition you selected for installation passes the S.M.A.R.T. monitor test (smartctl) but the tests indicate it will have a higher than average failure rate in the upcoming year.

Do you want to continue?"

Perhaps the wording that makes the distinction between the disk and the partition could be better. It should also be used in the failing message.

[Adrian]

Good suggestions, I changed it to what you proposed. Hope anticapitalista will have time to rebuild...

[Gordon Cooper]

(From an outsider (who has studied statistics and a bit of probability) ) I consider that "Do you want to continue" reads much better
than "Do you want to abort". Also I read the message as referring to the whole disk, not just the partition.

[tascoast]

Someone new to linux may panic with such a warning. I personally was surprised to hear my disc was about to go. I have the false positive however it now appears.

I'd consider the panic and potential (unnecessary perhaps even) costs a new user might experience after such a warning and if MX-14 should risk precipitating this scenario with such a live utility?

I'm all OK with the outcome myself but simply pose the above question to consider.

Kindly

Mick

[Adrian]

That's why we are fixing the code.

[BitJam]

@tascoast, what did you think of this message:

The disk with the partition you selected for installation passes the S.M.A.R.T. monitor test (smartctl) but the tests indicate it will have a higher than average failure rate in the upcoming year.

Do you want to continue?

This is greatly toned down from the message you got. As Adrian said, we are responding to the experience you had by reducing false positives and by scaling the message way back. In fact, this message understates the case a little since on average roughly half of all drive will have a failure rate higher than average (except in Lake Wobegon where all drives have a failure rate below average).

The other message about the disk failing will only appear if the smartmon tools say the disk is failing. Unlike your situation, this really is an emergency and no one should install on such a disk even though we will let people do so if they really want.

Again, this thread is our way of responding to the negative experience you had and is part of our attempt to prevent other people from having a similar negative experience. I am very much interested in your opinion but your comment is a little frustrating/confusing because we are working hard to correct the problem while your comment seems to ignore these efforts.