I have a small database on a local install of InfluxDB v1, less than 50 MB on disk.
But I can’t delete data. The query just hangs “forever”. (“Forever” being at least a week before I killed the processes that were piling up.)
My code does a delete from ... where multiple times a day and that hangs. If I go in with a blunt drop measurement that also hangs. This is not the first time this has happened, the last time a drop / create database did fix it, but apparently it comes back…
I don’t see any log activity to suggest anything is amiss, I don’t see anything odd in terms of cpu or memory usage or whatever while this is going on, the permissions on the data dir seem fine too. The query just hangs.
This was on a running 1.12.4 instance, on ubuntu 24.04. I tried updating influxdb to 1.13 just now, and now it works fine. Stupid in hindsight as now I don’t know whether it’s the reboot or the upgrade that fixed it.
Does anyone have any idea what’s going on?
Note: My username is ‘influxdata1’ because apparently this forum has a hard unchangeable link between your email address localpart and your username, and I use specific localparts for specific sites. I have nothing to do with InfluxData other than using their products.
The bugfixes in 1.13 are plausibly be the resolution here - though it’s hard to say for certain without getting way into the weeds, which is something we don’t need to do if it’s now happily working. Glad you got the resolution you needed, though!
Deletion is always slow in Influxdb 1 & 2. We’ve sped it up in 1.12.4 and a little more in 1.13.1, but DELETEs are always serialized, they touch a lot of series and index files, and they are better done with retention policies if your use case permits.
Folks - we’d love minimal repro cases with data, if you can provide the, For instance, @influxdata1 said their database was under 50MB. If it’s not proprietary data, a copy of that database and exact reproduction instructions would help us to understand what’s going on. Opening a github issue and assigning to me, davidby-influx, would be the best way to converse if you can provide us with a reproduction.
I don’t have a reproduction unfortunately. It is proprietary (customer) data so that’s sensitive, and it only happens very incidentally.
It’s not a matter of “slow deletes”. We’re talking days of hanging when left alone, and not loads of data across a lot of shards either, just stuff from the previous day.
This isn’t a lot to go on, I know. I mostly posted it to see if others would turn up. Which they did.
Do you have an idea for what checks I could run if it does happen again?
