a) i fixed the 'message xxxx' errors - they were indeed from an
accidentally preforked handle. i'll try to make a note or something
for apache docs this evening.
b)
I did a bunch of tracing this morning it seems that apache::dbi isn't
always issuing the rollback at cleanup
I set a trace level on my dbi arg, and have a handler print a marker
for the cleanup phase
the bulk of time I get something like this:
==============CLEANUP DBI::db=HASH(0x1a2c918) trace level set to
0x0/1 (DBI @ 0x0/0) in DBI 1.50-ithread (pid 24285)
<> FETCH('TraceLevel')= 1 ('TraceLevel' from cache) at DBI.pm
line 667
<- STORE('Username' 'XXXXX')= 1 at DBI.pm line 667
<> FETCH('Username')= 'XXXXX' ('Username' from cache) at DBI.pm
line 667
<- STORE('dbi_connect_closure' CODE(0x1a2be3c))= 1 at DBI.pm
line 683
if there's an invalid sql command executed in my transaction though,
==============CLEANUP
is all i get. no cleanup is issued (rollback/closure/etc) on the
dbi, and i'm still using a bad cached handle on subsequent page
views , necessitating a page start.
the above arguments were with a TraceLevel1 - i also did a TraceLevel
15, and saw no additional commands executed.
anyone have a suggestion on where I can look next?
