[gpfsug-discuss] Meaning of new fields in Watchfolder output

Hannappel, Juergen juergen.hannappel at desy.de
Mon Mar 21 12:37:39 GMT 2022


Hi,
for a test I read from a file with dd:

dd if=someFile bs=64k of=/dev/null
This will do 2 reads with 64k bytes, one read with 24053 and one final read with 0 bytes.

I now check the WF json output, looking for the new ReadOffset and bytesRead fields I get:
 kafkacat -b my-broker -t test1-watch | while read mist; do echo $mist | python -m json.tool | grep -i read ; done 

   "bytesRead": "0",
    "minReadOffset": "9223372036854775807",
    "maxReadOffset": "0",
That is the open()...

    "bytesRead": "65536",
    "minReadOffset": "0",
    "maxReadOffset": "65535",
That is the first read()....

    "bytesRead": "131072",
    "minReadOffset": "0",
    "maxReadOffset": "131071",
WHo reads 2x64k bytes from the beginning? Is the the accumulated read data?

    "bytesRead": "155125",
    "minReadOffset": "0",
    "maxReadOffset": "155124",
And again, now it looks as if in one event the entire file is read?
The documentation (https://www.ibm.com/docs/en/spectrum-scale/5.1.3?topic=folder-json-attributes-in-clustered-watch)
doesn't explain this....


-- 
Dr. Jürgen Hannappel  DESY/IT    Tel.  : +49 40 8998-4616




More information about the gpfsug-discuss mailing list