Обсуждение: PostgreSQL crashed, whole PC not responding
Hi, I'm upgrading to PostgreSQL 9.5.1 in 64-bit ArchLinux (Linux 4.4.1-2-ARCH) for around 4 days (previously we use 9.4.x), and the PostgreSQL crashed and make the whole PC down, with this last log from journalctl:
Mar 09 23:31:17 asd kernel: BUG: Bad page state in process postgres pfn:d18515
Mar 09 23:31:17 asd kernel: page:ffffea0034614540 count:0 mapcount:0 mapping: (null) index:0x1
Mar 09 23:31:17 asd kernel: flags: 0x2fffe0000000000()
Mar 09 23:31:17 asd kernel: page dumped because: page still charged to cgroup
Mar 09 23:31:17 asd kernel: page->mem_cgroup:0000000000020000
Mar 09 23:31:17 asd kernel: page:ffffea0034614540 count:0 mapcount:0 mapping: (null) index:0x1
Mar 09 23:31:17 asd kernel: flags: 0x2fffe0000000000()
Mar 09 23:31:17 asd kernel: page dumped because: page still charged to cgroup
Mar 09 23:31:17 asd kernel: page->mem_cgroup:0000000000020000
Is there any possible cause of this? or how to search the causation of this? since the only log i have is journalctl and log_min_duration_statement 2000 (and it shows nothing).
--
Best Regards,
Kiswono P / SoftDev
Kiswono P / SoftDev
GB
On Thu, Mar 10, 2016 at 2:37 AM, Kiswono Prayogo <kiswono@gmail.com> wrote:
Hi, I'm upgrading to PostgreSQL 9.5.1 in 64-bit ArchLinux (Linux 4.4.1-2-ARCH) for around 4 days (previously we use 9.4.x), and the PostgreSQL crashed and make the whole PC down, with this last log from journalctl:Mar 09 23:31:17 asd kernel: BUG: Bad page state in process postgres pfn:d18515
Mar 09 23:31:17 asd kernel: page:ffffea0034614540 count:0 mapcount:0 mapping: (null) index:0x1
Mar 09 23:31:17 asd kernel: flags: 0x2fffe0000000000()
Mar 09 23:31:17 asd kernel: page dumped because: page still charged to cgroup
Mar 09 23:31:17 asd kernel: page->mem_cgroup:0000000000020000
Is there any possible cause of this? or how to search the causation of this? since the only log i have is journalctl and log_min_duration_statement 2000 (and it shows nothing).
That looks like a kernel issue/bug. Both because the log comes from the kernel, and because PostgeSQL can't actually take down the whole PC as long as the kernel is working properly.