23
votes
InfluxDB has apparently shut down - and deleted! - two of its data centers and some customers did not get any warning
Link information
This data is scraped automatically and may be incorrect.
- Title
- Getting weird results from GCP europe-west1
- Authors
- Thijs, Fred334, PradoEagle, Jerry1333, Metropolis, mandelllee, Ping-Lin_Chang, Enlighten, pavbyte, Jay_Clifford, Julienfastre, matt-at-sparkmate, Oleh_Paduchak, RoyalBlock, Viktor_ingemarsson, Pooh, bcix3
- Published
- Jul 8 2023
- Word count
- 1102 words
From reading the thread and some other news I can find, it seems InfluxDB, the popular time-series database, has axed its
ap-southeast-2andeurope-west1data centers including their data.And as it then turns out, all they did was send three warning emails a few weeks apart each time. So if a customer had either turned all emails off, or for whatever reason filtered them out due to the marketing communication also arriving, they had zero warning their data is about to be removed.
This immediately raises two questions for me:
Feels really unprofessional, in particular for a paid service. Although, to be fair, I have a pretty one-sided picture so far, I have not yet talked to friends who work with InfluxDB.
I mean, dealing with issues like this internally....
People don't read their emails. We could send 50 warnings over two years that we're decommissioning a system, move your data. And we could have a 6 month 'offline but not deleted window' to account for the stragglers. And we'd still get people complaining we gave them no notice and we're responsible for their data loss 2 years later.
I'm sorry, if its a vendor you have important dealings with, you need to whitelist and label their domain. You need to read every email they send you. Otherwise stuff like this happens.
I deal with stuff like this routinely on a smaller scale. We know people don't read their emails and we make accommodations.
Sure, there are stragglers who won't notice for years, but in my experience the ones who don't read their emails generally notice pretty quickly when their account/service stops working. We give a year between account removal and data removal to account for seasonal usage and that's worked well for us.
Terminating a service and immediately removing all of the service's data without the possibility of recovery seems highly irresponsible to me. In my case, even if we somehow needed to remove a service and its data at the same time, we'd archive all of the data after service termination before deleting it off the instance, just in case something ends up being needed later. (And if a final archive isn't feasible for some reason, I would question whether sufficient regular backups were being done, which is whole separate red flag.)
Reading through the thread, it seems like emails simply weren't received if the end user opted out of "Marketing/Other" emails in their account settings. That's a bit different from blocking the address that marketing emails are coming from. Multiple people claim to have searched their inboxes and found no emails from Influx on the days they sent them.
On one hand i agree, we gave teams a year to move off, BUT we still held the data afterwards, just disabled accounts and turned off apps. Kept DBs and storage for a month.
Yea I probably should mention I'm not saying IndexDB did well on this. But I also feel for the flip side.
We recently had a similar surge of new customers from FutureQuest:
They've been stuck on their own unmaintainable in house control panel and finally conceded that it's not viable without pricing yourself out of customers.