get_psutil_disk_io_total(), ERROR: ValueError: not sure how to interpret line

[HangDazon]

Hello, I'm getting error when I add my server for testing

below are the ERROR Log:

2025-02-07 16:28:36.018 [ERROR] [sql:select /* pgwatch_generated */
  (extract(epoch from now()) * 1e9)::int8 as epoch_ns,
  read_count,
  write_count,
  read_bytes,
  write_bytes
from
  get_psutil_disk_io_total()
] [args:[]] [err:ERROR: ValueError: not sure how to interpret line '   8      16 sdb 4942 36 163044 9945 15868126 2066578 284609120 5227187 0 15401450 5237133 0 0 0 0\n' (SQLSTATE 38000)] [pid:3143844] [time:1.226035ms] Query

[pashagolub]

Timescale schema doesn't use partitions.

Next time, please, create a separate issues for different topics

[pashagolub]

"psutil" is known to behave differently depending on the used version and operating system, so if getting errors please adjust to your needs. "psutil" documentation here: https://psutil.readthedocs.io/en/latest/

You need to adjust this function:

            CREATE OR REPLACE FUNCTION get_psutil_disk_io_total(
            	OUT read_count float8, OUT write_count float8, OUT read_bytes float8, OUT write_bytes float8
            )
             LANGUAGE plpython3u
            AS $FUNCTION$
            from psutil import disk_io_counters
            dc = disk_io_counters(perdisk=False)
            if dc:
                return dc.read_count, dc.write_count, dc.read_bytes, dc.write_bytes
            else:
                return None, None, None, None
            $FUNCTION$;

[HangDazon]

@pashagolub , May I know which place to adjust, because I'm new to it.

I'm using on Redhat 8 with package: python3-psutil.x86_64 version:5.4.3-11.el8

thanks before hand.

[pashagolub]

get_psutil_disk_io_total() function is a function on a monitored server, which uses system psutil to get system information. So you need to debug things on your monitored system with Python. And then adjust the definition of database function

[HangDazon]

now working fine after modified the function, thanks for your help.

CREATE OR REPLACE FUNCTION public.get_psutil_disk_io_total(
    OUT read_count double precision, 
    OUT write_count double precision, 
    OUT read_bytes double precision, 
    OUT write_bytes double precision
) RETURNS record 
LANGUAGE plpython3u 
AS $$
try:
    # Read /proc/diskstats
    with open("/proc/diskstats", "r") as f:
        lines = f.readlines()

    # Initialize counters
    total_read_count = 0
    total_write_count = 0
    total_read_bytes = 0
    total_write_bytes = 0

    for line in lines:
        parts = line.split()
        
        # Ensure the line has at least 14 columns
        if len(parts) >= 14:
            # Extract relevant values
            read_count = int(parts[3])    # Field 4: Read I/O count
            read_bytes = int(parts[5]) * 512  # Field 6: Read sectors * 512 (sector size)
            write_count = int(parts[7])   # Field 8: Write I/O count
            write_bytes = int(parts[9]) * 512  # Field 10: Write sectors * 512

            # Accumulate values
            total_read_count += read_count
            total_write_count += write_count
            total_read_bytes += read_bytes
            total_write_bytes += write_bytes

    return (total_read_count, total_write_count, total_read_bytes, total_write_bytes)

except Exception as e:
    return (0, 0, 0, 0)  # Return 0s if any error occurs
$$;

[pashagolub]

Cool! What was the issue?

[HangDazon]

The default function is not working on Redhat8 OS. So, I just modified function like above, so it working fine.