Обсуждение: Log Clutter
Incidentally, am I the only one that finds the new 7.4 log messages too cluttered? In 7.3 my application printed progress messages using "raise notice" like this: 2004-08-20 11:58:16 [4369] NOTICE: Processing 6561 1/15127... 2004-08-20 11:58:19 [4369] NOTICE: Processing 6561 1/15127...Done Found 3800 2004-08-20 11:58:19 [4369] NOTICE: Processing 6794 2/15127... 2004-08-20 11:58:20 [4369] NOTICE: Processing 6794 2/15127...Done Found 4000 2004-08-20 11:58:20 [4369] NOTICE: Processing 10045 3/15127... 2004-08-20 11:58:20 [4369] NOTICE: Processing 10045 3/15127...Done Found 0 2004-08-20 11:58:20 [4369] NOTICE: Processing 12325 4/15127... 2004-08-20 11:58:20 [4369] NOTICE: Processing 12325 4/15127...Done Found 0 2004-08-20 11:58:20 [4369] NOTICE: Processing 14957 5/15127... 2004-08-20 11:58:20 [4369] NOTICE: Processing 14957 5/15127...Done Found 0 In 7.4 it helpfully tells me what function and line number is executing, but since it's always the same it's pretty useless: 2004-08-20 11:58:16 [4369] NOTICE: Processing 6561 1/15127... CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:19 [4369] NOTICE: Processing 6561 1/15127...Done Found 3800 CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:19 [4369] NOTICE: Processing 6794 2/15127... CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 6794 2/15127...Done Found 4000 CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 10045 3/15127... CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 10045 3/15127...Done Found 0 CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 12325 4/15127... CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 12325 4/15127...Done Found 0 CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 14957 5/15127... CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); 2004-08-20 11:58:20 [4369] NOTICE: Processing 14957 5/15127...Done Found 0 CONTEXT: PL/pgSQL function "process_regions" line 2 at return STATEMENT: select process_regions(); It also seems to print these after almost every query I have. That's because I use SQL functions lang_en()/lang_fr() to handle l10n. It also seems too chatty. 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining 2004-08-18 21:25:50 [16741] LOG: statement: select $1 [1] CONTEXT: SQL function "lang_en" during inlining As far as I can see there's no way to disable either of these two cases without disabling a lot more messages along with them. It would be nice... -- greg
Greg Stark <gsstark@mit.edu> writes: > Incidentally, am I the only one that finds the new 7.4 log messages too > cluttered? In 7.3 my application printed progress messages using "raise > notice" like this: We could have RAISE suppress all context. (I think at the moment it only suppresses the first level of context, which is probably a compromise that doesn't satisfy anybody.) I'd be inclined to do that only for the non-error forms of RAISE, though. regards, tom lane
On Fri, Aug 20, 2004 at 12:08:45PM -0400, Greg Stark wrote: > > Incidentally, am I the only one that finds the new 7.4 log messages too > cluttered? In 7.3 my application printed progress messages using "raise > notice" like this: You could execute a SET LOCAL log_error_verbosity within the function to decrease the verbosity ... the problem is that only a superuser can do that, but maybe using a SECURITY DEFINER function to do it for you helps. -- Alvaro Herrera (<alvherre[a]dcc.uchile.cl>) "Endurecerse, pero jamás perder la ternura" (E. Guevara)