Closed netrusov closed 14 hours ago
I see that 16.6 and 17.2 were released today, I'll wait for the new packages to be available and will test again.
edit: I just saw that there's a warning in the readme that mentions these specific versions of PostgreSQL. Sorry.
Don't think you need to be sorry here @netrusov
I also saw the hotfix from postgres today and immediately upgraded even though I didn't have issues with 16.5 like you did.
If anything, timescale probably needs to warn its users not to use pg16.5 and go straight to 16.6 (although it's not their fault in the first place).
I see that 16.6 and 17.2 were released today, I'll wait for the new packages to be available and will test again.
Yes, there were an unfortunate ABI change in the PostgreSQL upstreams package that affected us. You can see the entire thread to understand what happened.
It is possible for us to do some more work to avoid the risk of these kinds of changes affecting us, but this is not where we are right now.
edit: I just saw that there's a warning in the readme that mentions these specific versions of PostgreSQL. Sorry.
No need to be sorry here, it is better to raise one issue to many than one too few.
Worth adding is that because of the nature of the upstreams fix you will have problems upgrading from 17.1 to 17.2 (and similarly for the other versions), so you just skip the intermediate versions entirely.
Closing this as resolved for now. Please reopen if you think there is something more that we need to deal with.
What type of bug is this?
Crash
What subsystems and features are affected?
Other
What happened?
We've encountered an issue while upgrading our PostgreSQL from 16.4 to 17.1 We are using TimescaleDB extension version 2.17.2
After upgrade, we saw that some queries were causing an "Segmentation fault" exception. I've been experimenting with the data and came up with a reproducible test case, which is failing on PG 16.5 and 17.1, but everything is good on 16.4 and 17.0 Problem occurs when there are multiple chunks being used, I've reflected it in my test case.
Sometimes we see
ERROR: targetColnos does not match subplan target list
error, but after running query again it will cause segfault.TimescaleDB version affected
2.17.2
PostgreSQL version used
16.5, 17.1
What operating system did you use?
OEL, Debian/Ubuntu
What installation method did you use?
Deb/Apt, RPM
What platform did you run on?
On prem/Self-hosted
Relevant log output and stack trace