GroveStreams
jonrichings private msg quote post Address this user
I just tried to delete a few anomalous data points from the component/stream view. While the confirmation dialogue gave me confidence that this was possible and the data is removed from the table, it does not actually appear to be deleted from the store as a refresh brings it all back again. Am I missing something?

Also, just a thought for the future: being able to mark several data points for deletion at once would reduce the headache of having to delete a large range one at a time.
Post 1 IP   flag post
MikeMills private msg quote post Address this user
Delete should delete the samples. Just tried to reproduce this and couldn't. Are you sure it is not?

Also, deleting samples will not trigger a rollup recalc so the rollups will remain the same. This is by design. A design we regret and do plan on changing. Maybe this is what you're seeing? See this thread for more about delete and rollups.
Post 2 IP   flag post
jonrichings private msg quote post Address this user
Just triple checked. As far as I can see I am deleting the actual samples (*none* in the selection dialogue). Each sample appeared to delete from the list but comes back as soon as it is refreshed. If you still have access to our account take a look at compId: b8:27:eb:0a:f8:89, stream: da1, about 9pm on the 14 April. You'll see a big drop. This is barometric pressure so definitely anomalous (I need to find out where it came from). There is quite a few data points in this dip and it took quite a while to 'delete' them all, one at a time, twice. But it's all still there.

The rollups not recalculating automatically I can live with but the option to automatically recalc would be good.
Post 3 IP   flag post
MikeMills private msg quote post Address this user
I copied your stream and reproduced the issue. We'll dig into it. This one could take a while since we haven't reproduced it yet on one of our streams.
Post 4 IP   flag post
jonrichings private msg quote post Address this user
Phew, at least I'm not insane then!
Post 5 IP   flag post
PigFarmer private msg quote post Address this user
I have noticed the same thing. Delete a data point but it still shows up in my widget. Only way to get rid of it is to delete a range of data using the "delete range of data" function for the stream. Sounds like its found and looking into.
Post 6 IP   flag post
seckford private msg quote post Address this user
I've found that it's the other way around - I can't delete a range of data successfully, but individual deletes work. (And of course any rollups have to be deleted individually as well.)

Clearly there's some sort of deletion bug (and when deleting, F5 is your friend).

Will
Post 7 IP   flag post
MikeMills private msg quote post Address this user
Definitely a bug. We've reproduced it on a dev box and are working on a fix.
Post 8 IP   flag post
MikeMills private msg quote post Address this user
Patch applied. You may need to go back and delete the rogue samples again.

There was an issue with deleting samples that were over a day old and not located towards the tail of a regular stream. "Older" samples are stored in blocks for performance and scalability reasons. A second block is created during the deleting of samples within the block and then the original block is deleted. The bug was that the original block never got deleted which caused deleted samples to still appear and other samples to have duplicates appear.
Post 9 IP   flag post
jonrichings private msg quote post Address this user
That works Mike but I think there may be one or two gremlins left: out of curiosity I tried to edit a few values and add one... neither took. Maybe more symptoms of the same problem?
Post 10 IP   flag post
MikeMills private msg quote post Address this user
@jonrichings I just did a quick test and adding/editing worked for me. Are your values at the head (latest) of the stream, middle, or tail?
Post 11 IP   flag post
jonrichings private msg quote post Address this user
About where I was trying to delete them before, probably early-middle/tail.
Post 12 IP   flag post
MikeMills private msg quote post Address this user
@jonrichings I looked at your stream. The really low values are gone. Rollups are still not being recalced. Are you sure you're not looking at the rollups and thinking the lows are still there?

To force a rollup recalc, change the stream type to float, save the component, wait for the reconcile to finish, change the type back to double.

If the issues is still there, post the day and time so that I know which sample you're trying to change.
Post 13 IP   flag post
jonrichings private msg quote post Address this user
Mike, yes, the deletes worked just fine now. I deleted a whole bunch, but left two extremes high and low events. Those I tried to edit the values. I also tried to add a data point at 9pm exactly, but that disappeared on refresh. Have a look for yourself, on the 14th at between about 8pm and 10pm. Feels like I should be doing something to commit the changes though, am I missing something in the process or is it sufficient to just enter values and hit return?
Post 14 IP   flag post
MikeMills private msg quote post Address this user
There's a save button above the grid. You need to click that.
Post 15 IP   flag post
jonrichings private msg quote post Address this user
Ahh, silly me. I took that for an export type save since it is with the feed and excel buttons. Sorry 'bout that! But maybe you should rethink the position and icon? Just incase another user is as dumb as me.
Post 16 IP   flag post
jonrichings private msg quote post Address this user
It's also a little counterintuitive when the delete doesn't need saving (maybe it should).

And while I'm thinking about this area, it would be nice if you could click on the chart and have that fix highlighted table row so it is easier to find the offending point to delete or modify without it then scrolling away as you move the mouse to the table. Just something to add to the bottom of the "it would be nice if..." list.
Post 17 IP   flag post
jonrichings private msg quote post Address this user
Oh, and deleting, it would be hugely nice if you could select multiple rows (by CTRL-click maybe) before hitting a delete button (and then maybe save?). More consistent and far easier than deleting and confirming every freak datapoint. Yes, I know I can delete ranges but doing it from the table/chart view saves me from worrying about what the range(s) are.
Post 18 IP   flag post
2889 18 18
Log in or sign up to compose a reply.