Scheduled parity check notifications should be in a different notification category

[nmills3]

Currently when a scheduled parity check starts it displays a notification in the warning category. This isn't really a warning of anything though as it's just a scheduled check. It would probably make more sense for it to be in the notices category as the idea of the notification seems to be to notify you that the parity check is starting

[itimpi]

33 minutes ago, nmills3 said:

Currently when a scheduled parity check starts it displays a notification in the warning category. This isn't really a warning of anything though as it's just a scheduled check. It would probably make more sense for it to be in the notices category as the idea of the notification seems to be to notify you that the parity check is starting

At one level I agree with you, but since a parity check is a major array operation that has a long elapsed time and can significantly affect performance while it is running I can see why it might be deemed more appropriate to treat it as a warning rather than a simple notification.

[nmills3]

so for me personally, i think of warnings as a "Something has gone wrong but we can handle it without intervention", errors as "something has gone wrong and you need to fix it" and notices as a "just so you know, this is happening." I think the issue is that it's almost between the 2 categories.

 

I can see it being a warning when it's unexpected like the array wasn't shut down correctly and parity needs to be rebuilt or when a drive is replaced. But for schedules ones i have to disagree. As it's a scheduled event, the user should already be aware of it and should be expecting the performance hit anyway and most likely will have already set the check to run at a time where that is not an issue anyway. I think that scheduled checks should be considered notices and non scheduled checks should be warnings.

 

Also in response to your comment about it being a major array operation i have a question to ask. Given that the mover could also be considered a major array operation depending on how many files need to be moved, should it not also have a warning level norification when it starts? Currently it has the potential to also have a large performance impact on the array but doesn't have any notifications of any type from what I can find.

[itimpi]

35 minutes ago, nmills3 said:

Given that the mover could also be considered a major array operation depending on how many files need to be moved, should it not also have a warning level norification when it starts?

Since mover is about individual files rather than the whole array I would not consider it an ‘array level’ activity.   That does not mean that there is not a case for having more visibility of when it is running.
 

As you say mover can affect performance and if you have the Parity Check Tuning plugin installed you can get ‘notice’ level notifications when mover starts and stops if it happens at the same time an array operation is running as these two types of operation interact badly from a performance perspective.  
 

There is a trade-off around when to generate notifications.    I think many people prefer to get none at all in normal daily running if everything is running smoothly, and only get informed about deviations from that.

[trurl]

46 minutes ago, nmills3 said:

warnings as a "Something has gone wrong but we can handle it without intervention", errors as "something has gone wrong and you need to fix it"

Errors are displayed in different ways. I/O errors will show up in the Errors column on Main, but often these are handled without intervention, though they may mean something needs to be investigated.

 

SMART warnings are displayed in the Dashboard as 👎and depending on the specific SMART attribute, might mean something needs fixing, and Unraid definitely can't fix these warnings without intervention.

[nmills3]

3 minutes ago, trurl said:

I/O errors will show up in the Errors column on Main

don't some IO related errors show as warnings instead of errors? for example, i'm pretty sure if a drives "UDMA CRC error count" increases, that shows as a warning to the user and not an error in the notifications in the top right. Then once the error count gets high enough the disk gets disabled and then an actual error notification is sent. Hence my example of a warning being an event that doesn't need user intervention. Then once it gets to the point where the drive is disabled, then you get the error notification because at that point it falls into the "you need to fix it" category.

 

I would also like to be clear that for the purpose of warnings, errors and notices. I'm only talking about the notifications in the top left on the UI

[nmills3]

I also don't think it's helping that since the recent change where notifications are now hidden in the little drop down in the top left, you are now expected to go off the colour of the bell to know what notifications are waiting for you. If i see a green bell i assume something is there i can just ignore. A yellow bell means it's something i need to look at incase a larger problem is starting and then a red bell means something has gone very wrong. So everytime a parity check happens, i get the yellow bell and think something is going wrong, then it just turns out to be a parity check notification that could have been ignored

[trurl]

5 minutes ago, nmills3 said:

if a drives "UDMA CRC error count" increases, that shows as a warning to the user and not an error in the notifications in the top right. Then once the error count gets high enough the disk gets disabled

A disk only gets disabled when a write to it fails.

 

Better if you setup Notifications to alert you immediately by email or other agent instead of waiting until you check the webUI. I don't get any Notifications in the webUI.