Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-08-09 Thread Justin Workman
On Tue, Aug 8, 2017 at 8:17 PM, Tom Lane wrote: > Peter Geoghegan writes: > > On Thu, Jan 19, 2017 at 5:45 PM, Peter Geoghegan wrote: > >> A customer is on 9.6.1, and complains of a segfault observed at least > >> 3 times. > > ... > > For the

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-08-08 Thread Tom Lane
Peter Geoghegan writes: > On Thu, Jan 19, 2017 at 5:45 PM, Peter Geoghegan wrote: >> A customer is on 9.6.1, and complains of a segfault observed at least >> 3 times. > ... > For the sake of the archives: this now looks very much like the issue > that Tom just

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-08-08 Thread Peter Geoghegan
On Thu, Jan 19, 2017 at 5:45 PM, Peter Geoghegan wrote: > A customer is on 9.6.1, and complains of a segfault observed at least > 3 times. > I can use GDB to get details of the instruction pointer that appeared > in the kernel trap error, which shows a function from the expanded

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-02-16 Thread Amit Kapila
On Thu, Feb 16, 2017 at 11:45 AM, Justin Workman wrote: >> It would help to know the data types of the columns involved in this >> query; but just eyeballing it, it doesn't look like it involves any >> array operations, so it's pretty hard to believe that the expanded-object

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-02-16 Thread Justin Workman
> > It would help to know the data types of the columns involved in this > query; but just eyeballing it, it doesn't look like it involves any > array operations, so it's pretty hard to believe that the expanded-object > code could have gotten invoked intentionally. (The mere presence of > an

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-02-03 Thread Peter Geoghegan
On Thu, Jan 19, 2017 at 6:39 PM, Tom Lane wrote: > If $customer wants a quick fix, I'd suggest seeing whether disabling > parallel query makes the problem go away. That might be a good first > step anyway, just to narrow down where the problem lies. I no longer work at

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-01-20 Thread Robert Haas
On Thu, Jan 19, 2017 at 9:39 PM, Tom Lane wrote: > If I had to bet on the basis of this much info, I would bet that the > parallel-query infrastructure is dropping the ball somewhere and > transmitting a corrupted datum that accidentally looks like it is > an expanded-object

Re: [HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-01-19 Thread Tom Lane
Peter Geoghegan writes: > A customer is on 9.6.1, and complains of a segfault observed at least > 3 times. In all cases, the logs look like this: > ... > I can use GDB to get details of the instruction pointer that appeared > in the kernel trap error, which shows a function from

[HACKERS] Possible issue with expanded object infrastructure on Postgres 9.6.1

2017-01-19 Thread Peter Geoghegan
A customer is on 9.6.1, and complains of a segfault observed at least 3 times. In all cases, the logs look like this: Jan 11 16:11:07 ip-10-0-118-82 kernel: [41913.530453] traps: postgres[6561] general protection ip:55fcf08b0491 sp:7ffc17dfa650 error:0 in postgres[55fcf0557000+638000] Jan 11