Subject | Re: [firebird-support] Ackward trace information |
---|---|
Author | Anderson Farias |
Post date | 2011-01-20T12:34:47Z |
Hi Thomas,
|Does this also happen with other tables or just the one above?
It happens to some other tables. And... I was able to actually narrow IN WHAT circunstances it happens (but still don't know WHY sice fb raises no exception)
If I do
SELECT * FROM SOME_TROUBLED_TABLE
and fetch, lets say first 100 records and commit. than... no FAILED shows up! BUT, if I before commiting, fetch the remaining records than, at some point will *fail* (FAILED entry is recorded to audit log file)
But why? Is my data or database broken? Why fb do not raise any exceptions?
|It would be interesting if you have been running a user trace
|or a system audit and with what trace configuration.
Yes, sorry. I'm running system audit.
Everything is *enabled* except:
log_context
log_blr_requests
print_blr
log_dyn_requests
print_dyn false
there's no filters (include or exclude), max_log_size is set to 0, time_threshold is set to 0, and max_sql_length is set to 30000.
BUT it seems its just I *my* huge mistake here:
1) Since max_log_size is set to 0 the file has grown over 500Gb and the text editor was not showing last entries (so, they may acctualy be there)
2) I started renaming the files to get a clean result on some tests. I did not stoped/started fb service, nor disconected/connected from client app. So, I guess fb audit is not Ok with that.
Regards,
Anderson
|Does this also happen with other tables or just the one above?
It happens to some other tables. And... I was able to actually narrow IN WHAT circunstances it happens (but still don't know WHY sice fb raises no exception)
If I do
SELECT * FROM SOME_TROUBLED_TABLE
and fetch, lets say first 100 records and commit. than... no FAILED shows up! BUT, if I before commiting, fetch the remaining records than, at some point will *fail* (FAILED entry is recorded to audit log file)
But why? Is my data or database broken? Why fb do not raise any exceptions?
|It would be interesting if you have been running a user trace
|or a system audit and with what trace configuration.
Yes, sorry. I'm running system audit.
Everything is *enabled* except:
log_context
log_blr_requests
print_blr
log_dyn_requests
print_dyn false
there's no filters (include or exclude), max_log_size is set to 0, time_threshold is set to 0, and max_sql_length is set to 30000.
BUT it seems its just I *my* huge mistake here:
1) Since max_log_size is set to 0 the file has grown over 500Gb and the text editor was not showing last entries (so, they may acctualy be there)
2) I started renaming the files to get a clean result on some tests. I did not stoped/started fb service, nor disconected/connected from client app. So, I guess fb audit is not Ok with that.
Regards,
Anderson