What does a kernel trace via truss/ktrace/strace/tusc/par/etc. show? That ought to confirm ikegami's suspicion of signalry right away.
That suspicion, FWIW, strikes me as probably correct. IIRC, at least some versions of the OCI handled at least SIGINTs themselves, and the Sys::SigAction pod documents one coder's woe in dealing with 5.8's signal behavior vis-a-vis Oracle. Your versions of Perl, DBD::Oracle and the OCI will probably factor in to what's gone awry, as might something subtly wrong with your connection to the Oracle data source.
-pilcrowIn reply to Re: DBI messes up sleep
by pilcrow
in thread DBI messes up sleep
by perljunky
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |