Currently it is hard to automatically parse mongod and mongos logfiles due to little inconsistencies and ambiguities that can easily be fixed.

numYields

Wed Sep 05 18:17:35 [conn106842] getmore test.docs query: { ps: 3 } cursorid:77621193264928 ntoreturn:0 keyUpdates:0 numYields: 7 locks(micros) r:12009 nreturned:1044 reslen:2628447 15ms

Most counters like ntoreturn:1, nreturned:1, ... have no space after the colon, but numYields: 1 does have a space after the colon, making its value a separate token when splitting the line on spaces. Please remove the space after numYields:.

Space before "ms"

Most timed operations print their duration as ###ms (#=digit). Some, however, have a space before the ms.

Wed Sep 05 18:56:05 [DataFileSync] FlushViewOfFile for F:/data/local.3 failed with error 1 after 1 attempts taking 933 ms

Please remove the space. As this is a counter like the ones above, it would also make sense to replace this with duration:### instead for consistency.

read/write lock syntax

The read and write lock microsecond numbers currently have this syntax:

Wed Jun 19 13:48:13.454 [conn2] insert test.docs ninserted:1 keyUpdates:0 locks(micros) w:231001 230ms

The w: is ambiguous with for example the getlasterror command with write concern. The actual indicator locks(micros) is a separate and not very useful token.

Suggestion: remove the locks(micros) and rename the actual w: and r: labels to wlock: and rlock:.

Server Restarted

  
  
  
***** SERVER RESTARTED *****
  
  

Instead of the two empty lines before and after the server restarted messages, it should be in one single line, with a date and time like all the other lines.

These are trivial changes but would be very useful for machine parsing.