Potential Sharing Issue - query sometimes empty sometimes returns data

I am using the paid shared infrastructure and ingested some data over the last few days.
Now, when querying the data, it flip-flops in and out of existence. (see video)

I am using the EU Shared Infrastructure if that makes a difference.
Which is reporting version: Dgraph Version v21.03.0-78-ge4ad0b113

The aggregate count is also not stable

Connecting via Ratel to inspect the cluster shows 1 alpha to be red,

I tried creating a support ticket, but creating tickets fails

We are taking a look at this issue.

1 Like

Hi @jcsrb

Can you please check now if you get consistent results for the queries (including the aggregate count) ?

I see all 3 alphas are green now,

but I am still seeing the odd results

I noticed the count seams to be different by 5, and the node I queried I suppose is one of the 5 “phantom” nodes – if I query for other document it is stable.

I will start a new shared backend and import the data into,
Would love to know how to avoid getting into this situation in the first place :slight_smile:

We have also seen “context deadline exceeded” – that has been documented on the forum before for the EU region – maybe during such an event the shards got out of sync.

Thanks for the detailed info. At this point, we’d like to take a look at your backend for further analysis.
Would you be able to open a support ticket with us ? I just tested creating a new ticket via the Cloud UI and it seems to work fine for me.
I could not find any ‘Shared’ backends using your email address (for Discuss). If you can provide us with your backend info., preferably on a ticket, that’d be helpful.

I exported the data from that cluster as described here and created a new cluster in shared on EU infrastructure (I have a requirement to use EU-based services)

On the new one (nameless-brook-360107.eu-central-1), counts are also unstable

I have sent a DM with the details, and how trying to create a ticket looks from my account

Update: I was added to Zendesk and was able to create a ticket there

Thanks for the ticket!
Let’s continue our discussions there.

Please keep us posted on this issue. Thanks

The current status is that after some time it stabilized, then I added some more data, and it started showing a discrepancy of 1 in the count.
After a time (unsure if there was an intervention) the second cluster started also showing consistent numbers, I have since removed that second cluster as it was for testing only.